# Running on cnode7-012 # Started at Thu Dec 5 16:32:59 CST 2024 # SLURMD_NODENAME=cnode7-012 # SLURM_CLUSTER_NAME=slurm # SLURM_CONF=/cm/shared/apps/slurm/var/etc/slurm/slurm.conf # SLURM_CPUS_ON_NODE=224 # SLURM_CPUS_PER_TASK=8 # SLURM_EXPORT_ENV=PATH # SLURM_GET_USER_ENV=1 # SLURM_GPUS_ON_NODE=8 # SLURM_GPUS_PER_NODE=8 # SLURM_GTIDS=0 # SLURM_JOBID=6492 # SLURM_JOB_CPUS_PER_NODE='224(x2)' # SLURM_JOB_END_TIME=1764923577 # SLURM_JOB_GID=1026 # SLURM_JOB_GPUS=0,1,2,3,4,5,6,7 # SLURM_JOB_ID=6492 # SLURM_JOB_NAME=exp_owsm/s2t_train_05b_ds_raw_bpe50000/train.log # SLURM_JOB_NODELIST='cnode7-[012-013]' # SLURM_JOB_NUM_NODES=2 # SLURM_JOB_PARTITION=p2 # SLURM_JOB_QOS=normal # SLURM_JOB_START_TIME=1733387577 # SLURM_JOB_UID=1026 # SLURM_JOB_USER=williamchen # SLURM_LOCALID=0 # SLURM_MEM_PER_NODE=2048000 # SLURM_NNODES=2 # SLURM_NODEID=0 # SLURM_NODELIST='cnode7-[012-013]' # SLURM_NODE_ALIASES='(null)' # SLURM_OPEN_MODE=a # SLURM_PRIO_PROCESS=0 # SLURM_PROCID=0 # SLURM_SUBMIT_DIR=/mnt/home/williamchen/espnet/egs2/owsm_v3.1/s2t1 # SLURM_SUBMIT_HOST=154-T2-P1-NVR # SLURM_TASKS_PER_NODE='28(x2)' # SLURM_TASK_PID=1271202 # SLURM_TOPOLOGY_ADDR=cnode7-012 # SLURM_TOPOLOGY_ADDR_PATTERN=node # SLURM_WORKING_CLUSTER=slurm:154-T2-P1-NVR:6817:9984:109 # srun --export=ALL python3 -m espnet2.bin.s2t_train --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp_owsm/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp_owsm/s2t_train_05b_ds_raw_bpe50000 --config conf/train_05b_ds.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp_owsm/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp,speech,kaldi_ark --train_shape_file exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape --fold_length 150 --train_data_path_and_name_and_type exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev,text_prev,text --fold_length 150 --train_data_path_and_name_and_type exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc,text_ctc,text --fold_length 150 --train_data_path_and_name_and_type exp_owsm/s2t_stats_raw_bpe50000/splits8/text,text,text --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --ngpu 8 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///mnt/home/williamchen/espnet/egs2/owsm_v3.1/s2t1/exp_owsm/s2t_train_05b_ds_raw_bpe50000/.dist_init_5a61425c-1365-44b0-8827-78539a6df709 [2024-12-05 16:33:12,744] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:33:13,821] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) /mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/bin/python3 /mnt/home/williamchen/espnet/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp_owsm/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp_owsm/s2t_train_05b_ds_raw_bpe50000 --config conf/train_05b_ds.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp_owsm/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp,speech,kaldi_ark --train_shape_file exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape --fold_length 150 --train_data_path_and_name_and_type exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev,text_prev,text --fold_length 150 --train_data_path_and_name_and_type exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc,text_ctc,text --fold_length 150 --train_data_path_and_name_and_type exp_owsm/s2t_stats_raw_bpe50000/splits8/text,text,text --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --ngpu 8 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///mnt/home/williamchen/espnet/egs2/owsm_v3.1/s2t1/exp_owsm/s2t_train_05b_ds_raw_bpe50000/.dist_init_5a61425c-1365-44b0-8827-78539a6df709 /mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/bin/python3 /mnt/home/williamchen/espnet/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp_owsm/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp_owsm/s2t_train_05b_ds_raw_bpe50000 --config conf/train_05b_ds.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp_owsm/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp,speech,kaldi_ark --train_shape_file exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape --fold_length 150 --train_data_path_and_name_and_type exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev,text_prev,text --fold_length 150 --train_data_path_and_name_and_type exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc,text_ctc,text --fold_length 150 --train_data_path_and_name_and_type exp_owsm/s2t_stats_raw_bpe50000/splits8/text,text,text --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --ngpu 8 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///mnt/home/williamchen/espnet/egs2/owsm_v3.1/s2t1/exp_owsm/s2t_train_05b_ds_raw_bpe50000/.dist_init_5a61425c-1365-44b0-8827-78539a6df709 [2024-12-05 16:33:32,842] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:33:33,274] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:33:33,598] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:33:33,687] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:33:33,887] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:33:34,052] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:33:34,072] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:33:34,086] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:33:35,209] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:33:35,234] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:33:35,640] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:33:35,709] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:33:35,750] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:33:35,850] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:33:35,964] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:33:35,978] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [W1205 16:33:40.249499608 Utils.hpp:164] Warning: Environment variable NCCL_BLOCKING_WAIT is deprecated; use TORCH_NCCL_BLOCKING_WAIT instead (function operator()) [W1205 16:33:41.693936888 Utils.hpp:164] Warning: Environment variable NCCL_BLOCKING_WAIT is deprecated; use TORCH_NCCL_BLOCKING_WAIT instead (function operator()) [W1205 16:33:41.094114010 Utils.hpp:164] Warning: Environment variable NCCL_BLOCKING_WAIT is deprecated; use TORCH_NCCL_BLOCKING_WAIT instead (function operator()) [cnode7-012:0/16] 2024-12-05 16:33:41,743 (s2t:462) INFO: Vocabulary size: 50002 [W1205 16:33:41.124905825 Utils.hpp:164] Warning: Environment variable NCCL_BLOCKING_WAIT is deprecated; use TORCH_NCCL_BLOCKING_WAIT instead (function operator()) [W1205 16:33:41.238554198 Utils.hpp:164] Warning: Environment variable NCCL_BLOCKING_WAIT is deprecated; use TORCH_NCCL_BLOCKING_WAIT instead (function operator()) [W1205 16:33:41.356129171 Utils.hpp:164] Warning: Environment variable NCCL_BLOCKING_WAIT is deprecated; use TORCH_NCCL_BLOCKING_WAIT instead (function operator()) [W1205 16:33:42.556111480 Utils.hpp:164] Warning: Environment variable NCCL_BLOCKING_WAIT is deprecated; use TORCH_NCCL_BLOCKING_WAIT instead (function operator()) [W1205 16:33:42.561568295 Utils.hpp:164] Warning: Environment variable NCCL_BLOCKING_WAIT is deprecated; use TORCH_NCCL_BLOCKING_WAIT instead (function operator()) [W1205 16:33:43.341781915 Utils.hpp:164] Warning: Environment variable NCCL_BLOCKING_WAIT is deprecated; use TORCH_NCCL_BLOCKING_WAIT instead (function operator()) [W1205 16:33:43.389135257 Utils.hpp:164] Warning: Environment variable NCCL_BLOCKING_WAIT is deprecated; use TORCH_NCCL_BLOCKING_WAIT instead (function operator()) [W1205 16:33:43.591531175 Utils.hpp:164] Warning: Environment variable NCCL_BLOCKING_WAIT is deprecated; use TORCH_NCCL_BLOCKING_WAIT instead (function operator()) [W1205 16:33:43.629085428 Utils.hpp:164] Warning: Environment variable NCCL_BLOCKING_WAIT is deprecated; use TORCH_NCCL_BLOCKING_WAIT instead (function operator()) [W1205 16:33:43.787734351 Utils.hpp:164] Warning: Environment variable NCCL_BLOCKING_WAIT is deprecated; use TORCH_NCCL_BLOCKING_WAIT instead (function operator()) [W1205 16:33:43.018435442 Utils.hpp:164] Warning: Environment variable NCCL_BLOCKING_WAIT is deprecated; use TORCH_NCCL_BLOCKING_WAIT instead (function operator()) [W1205 16:33:43.045873414 Utils.hpp:164] Warning: Environment variable NCCL_BLOCKING_WAIT is deprecated; use TORCH_NCCL_BLOCKING_WAIT instead (function operator()) [W1205 16:33:43.062686209 Utils.hpp:164] Warning: Environment variable NCCL_BLOCKING_WAIT is deprecated; use TORCH_NCCL_BLOCKING_WAIT instead (function operator()) [cnode7-012:0/16] 2024-12-05 16:33:45,684 (abs_task:1383) INFO: pytorch.version=2.4.0+cu121, cuda.available=True, cudnn.version=90100, cudnn.benchmark=False, cudnn.deterministic=True [cnode7-012:0/16] 2024-12-05 16:33:45,690 (abs_task:1384) INFO: Model structure: ESPnetS2TModel( (frontend): DefaultFrontend( (stft): Stft(n_fft=512, win_length=400, hop_length=160, center=True, normalized=False, onesided=True) (frontend): Frontend() (logmel): LogMel(sr=16000, n_fft=512, n_mels=80, fmin=0, fmax=8000.0, htk=False) ) (specaug): SpecAug( (freq_mask): MaskAlongAxis(mask_width_range=[0, 27], num_mask=2, axis=freq) (time_mask): MaskAlongAxisVariableMaxWidth(mask_width_ratio_range=[0.0, 0.05], num_mask=10, axis=time) ) (normalize): GlobalMVN(stats_file=exp_owsm/s2t_stats_raw_bpe50000/train/feats_stats.npz, norm_means=True, norm_vars=True) (encoder): TransformerEncoder( (embed): Conv2dSubsampling( (conv): Sequential( (0): Conv2d(1, 1024, kernel_size=(3, 3), stride=(2, 2)) (1): ReLU() (2): Conv2d(1024, 1024, kernel_size=(3, 3), stride=(2, 2)) (3): ReLU() ) (out): Sequential( (0): Linear(in_features=19456, out_features=1024, bias=True) (1): PositionalEncoding( (dropout): Dropout(p=0.1, inplace=False) ) ) ) (encoders): MultiSequential( (0): EncoderLayer( (self_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (feed_forward): PositionwiseFeedForward( (w_1): Linear(in_features=1024, out_features=4096, bias=True) (w_2): Linear(in_features=4096, out_features=1024, bias=True) (dropout): Dropout(p=0.1, inplace=False) (activation): ReLU() ) (norm1): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm2): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) (1): EncoderLayer( (self_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (feed_forward): PositionwiseFeedForward( (w_1): Linear(in_features=1024, out_features=4096, bias=True) (w_2): Linear(in_features=4096, out_features=1024, bias=True) (dropout): Dropout(p=0.1, inplace=False) (activation): ReLU() ) (norm1): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm2): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) (2): EncoderLayer( (self_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (feed_forward): PositionwiseFeedForward( (w_1): Linear(in_features=1024, out_features=4096, bias=True) (w_2): Linear(in_features=4096, out_features=1024, bias=True) (dropout): Dropout(p=0.1, inplace=False) (activation): ReLU() ) (norm1): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm2): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) (3): EncoderLayer( (self_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (feed_forward): PositionwiseFeedForward( (w_1): Linear(in_features=1024, out_features=4096, bias=True) (w_2): Linear(in_features=4096, out_features=1024, bias=True) (dropout): Dropout(p=0.1, inplace=False) (activation): ReLU() ) (norm1): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm2): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) (4): EncoderLayer( (self_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (feed_forward): PositionwiseFeedForward( (w_1): Linear(in_features=1024, out_features=4096, bias=True) (w_2): Linear(in_features=4096, out_features=1024, bias=True) (dropout): Dropout(p=0.1, inplace=False) (activation): ReLU() ) (norm1): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm2): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) (5): EncoderLayer( (self_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (feed_forward): PositionwiseFeedForward( (w_1): Linear(in_features=1024, out_features=4096, bias=True) (w_2): Linear(in_features=4096, out_features=1024, bias=True) (dropout): Dropout(p=0.1, inplace=False) (activation): ReLU() ) (norm1): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm2): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) (6): EncoderLayer( (self_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (feed_forward): PositionwiseFeedForward( (w_1): Linear(in_features=1024, out_features=4096, bias=True) (w_2): Linear(in_features=4096, out_features=1024, bias=True) (dropout): Dropout(p=0.1, inplace=False) (activation): ReLU() ) (norm1): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm2): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) (7): EncoderLayer( (self_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (feed_forward): PositionwiseFeedForward( (w_1): Linear(in_features=1024, out_features=4096, bias=True) (w_2): Linear(in_features=4096, out_features=1024, bias=True) (dropout): Dropout(p=0.1, inplace=False) (activation): ReLU() ) (norm1): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm2): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) (8): EncoderLayer( (self_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (feed_forward): PositionwiseFeedForward( (w_1): Linear(in_features=1024, out_features=4096, bias=True) (w_2): Linear(in_features=4096, out_features=1024, bias=True) (dropout): Dropout(p=0.1, inplace=False) (activation): ReLU() ) (norm1): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm2): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) (9): EncoderLayer( (self_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (feed_forward): PositionwiseFeedForward( (w_1): Linear(in_features=1024, out_features=4096, bias=True) (w_2): Linear(in_features=4096, out_features=1024, bias=True) (dropout): Dropout(p=0.1, inplace=False) (activation): ReLU() ) (norm1): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm2): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) (10): EncoderLayer( (self_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (feed_forward): PositionwiseFeedForward( (w_1): Linear(in_features=1024, out_features=4096, bias=True) (w_2): Linear(in_features=4096, out_features=1024, bias=True) (dropout): Dropout(p=0.1, inplace=False) (activation): ReLU() ) (norm1): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm2): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) (11): EncoderLayer( (self_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (feed_forward): PositionwiseFeedForward( (w_1): Linear(in_features=1024, out_features=4096, bias=True) (w_2): Linear(in_features=4096, out_features=1024, bias=True) (dropout): Dropout(p=0.1, inplace=False) (activation): ReLU() ) (norm1): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm2): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) (12): EncoderLayer( (self_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (feed_forward): PositionwiseFeedForward( (w_1): Linear(in_features=1024, out_features=4096, bias=True) (w_2): Linear(in_features=4096, out_features=1024, bias=True) (dropout): Dropout(p=0.1, inplace=False) (activation): ReLU() ) (norm1): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm2): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) (13): EncoderLayer( (self_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (feed_forward): PositionwiseFeedForward( (w_1): Linear(in_features=1024, out_features=4096, bias=True) (w_2): Linear(in_features=4096, out_features=1024, bias=True) (dropout): Dropout(p=0.1, inplace=False) (activation): ReLU() ) (norm1): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm2): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) (14): EncoderLayer( (self_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (feed_forward): PositionwiseFeedForward( (w_1): Linear(in_features=1024, out_features=4096, bias=True) (w_2): Linear(in_features=4096, out_features=1024, bias=True) (dropout): Dropout(p=0.1, inplace=False) (activation): ReLU() ) (norm1): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm2): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) (15): EncoderLayer( (self_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (feed_forward): PositionwiseFeedForward( (w_1): Linear(in_features=1024, out_features=4096, bias=True) (w_2): Linear(in_features=4096, out_features=1024, bias=True) (dropout): Dropout(p=0.1, inplace=False) (activation): ReLU() ) (norm1): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm2): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) ) (after_norm): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) ) (decoder): TransformerDecoder( (embed): Sequential( (0): Embedding(50002, 1024) (1): PositionalEncoding( (dropout): Dropout(p=0.1, inplace=False) ) ) (after_norm): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (output_layer): Linear(in_features=1024, out_features=50002, bias=True) (decoders): MultiSequential( (0): DecoderLayer( (self_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (src_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (feed_forward): PositionwiseFeedForward( (w_1): Linear(in_features=1024, out_features=4096, bias=True) (w_2): Linear(in_features=4096, out_features=1024, bias=True) (dropout): Dropout(p=0.1, inplace=False) (activation): ReLU() ) (norm1): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm2): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm3): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) (1): DecoderLayer( (self_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (src_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (feed_forward): PositionwiseFeedForward( (w_1): Linear(in_features=1024, out_features=4096, bias=True) (w_2): Linear(in_features=4096, out_features=1024, bias=True) (dropout): Dropout(p=0.1, inplace=False) (activation): ReLU() ) (norm1): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm2): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm3): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) (2): DecoderLayer( (self_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (src_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (feed_forward): PositionwiseFeedForward( (w_1): Linear(in_features=1024, out_features=4096, bias=True) (w_2): Linear(in_features=4096, out_features=1024, bias=True) (dropout): Dropout(p=0.1, inplace=False) (activation): ReLU() ) (norm1): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm2): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm3): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) (3): DecoderLayer( (self_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (src_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (feed_forward): PositionwiseFeedForward( (w_1): Linear(in_features=1024, out_features=4096, bias=True) (w_2): Linear(in_features=4096, out_features=1024, bias=True) (dropout): Dropout(p=0.1, inplace=False) (activation): ReLU() ) (norm1): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm2): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm3): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) (4): DecoderLayer( (self_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (src_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (feed_forward): PositionwiseFeedForward( (w_1): Linear(in_features=1024, out_features=4096, bias=True) (w_2): Linear(in_features=4096, out_features=1024, bias=True) (dropout): Dropout(p=0.1, inplace=False) (activation): ReLU() ) (norm1): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm2): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm3): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) (5): DecoderLayer( (self_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (src_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (feed_forward): PositionwiseFeedForward( (w_1): Linear(in_features=1024, out_features=4096, bias=True) (w_2): Linear(in_features=4096, out_features=1024, bias=True) (dropout): Dropout(p=0.1, inplace=False) (activation): ReLU() ) (norm1): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm2): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm3): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) (6): DecoderLayer( (self_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (src_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (feed_forward): PositionwiseFeedForward( (w_1): Linear(in_features=1024, out_features=4096, bias=True) (w_2): Linear(in_features=4096, out_features=1024, bias=True) (dropout): Dropout(p=0.1, inplace=False) (activation): ReLU() ) (norm1): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm2): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm3): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) (7): DecoderLayer( (self_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (src_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (feed_forward): PositionwiseFeedForward( (w_1): Linear(in_features=1024, out_features=4096, bias=True) (w_2): Linear(in_features=4096, out_features=1024, bias=True) (dropout): Dropout(p=0.1, inplace=False) (activation): ReLU() ) (norm1): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm2): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm3): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) (8): DecoderLayer( (self_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (src_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (feed_forward): PositionwiseFeedForward( (w_1): Linear(in_features=1024, out_features=4096, bias=True) (w_2): Linear(in_features=4096, out_features=1024, bias=True) (dropout): Dropout(p=0.1, inplace=False) (activation): ReLU() ) (norm1): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm2): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm3): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) (9): DecoderLayer( (self_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (src_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (feed_forward): PositionwiseFeedForward( (w_1): Linear(in_features=1024, out_features=4096, bias=True) (w_2): Linear(in_features=4096, out_features=1024, bias=True) (dropout): Dropout(p=0.1, inplace=False) (activation): ReLU() ) (norm1): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm2): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm3): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) (10): DecoderLayer( (self_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (src_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (feed_forward): PositionwiseFeedForward( (w_1): Linear(in_features=1024, out_features=4096, bias=True) (w_2): Linear(in_features=4096, out_features=1024, bias=True) (dropout): Dropout(p=0.1, inplace=False) (activation): ReLU() ) (norm1): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm2): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm3): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) (11): DecoderLayer( (self_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (src_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (feed_forward): PositionwiseFeedForward( (w_1): Linear(in_features=1024, out_features=4096, bias=True) (w_2): Linear(in_features=4096, out_features=1024, bias=True) (dropout): Dropout(p=0.1, inplace=False) (activation): ReLU() ) (norm1): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm2): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm3): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) (12): DecoderLayer( (self_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (src_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (feed_forward): PositionwiseFeedForward( (w_1): Linear(in_features=1024, out_features=4096, bias=True) (w_2): Linear(in_features=4096, out_features=1024, bias=True) (dropout): Dropout(p=0.1, inplace=False) (activation): ReLU() ) (norm1): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm2): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm3): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) (13): DecoderLayer( (self_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (src_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (feed_forward): PositionwiseFeedForward( (w_1): Linear(in_features=1024, out_features=4096, bias=True) (w_2): Linear(in_features=4096, out_features=1024, bias=True) (dropout): Dropout(p=0.1, inplace=False) (activation): ReLU() ) (norm1): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm2): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm3): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) (14): DecoderLayer( (self_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (src_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (feed_forward): PositionwiseFeedForward( (w_1): Linear(in_features=1024, out_features=4096, bias=True) (w_2): Linear(in_features=4096, out_features=1024, bias=True) (dropout): Dropout(p=0.1, inplace=False) (activation): ReLU() ) (norm1): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm2): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm3): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) (15): DecoderLayer( (self_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (src_attn): MultiHeadedAttention( (linear_q): Linear(in_features=1024, out_features=1024, bias=True) (linear_k): Linear(in_features=1024, out_features=1024, bias=True) (linear_v): Linear(in_features=1024, out_features=1024, bias=True) (linear_out): Linear(in_features=1024, out_features=1024, bias=True) (dropout): Identity() (q_norm): Identity() (k_norm): Identity() ) (feed_forward): PositionwiseFeedForward( (w_1): Linear(in_features=1024, out_features=4096, bias=True) (w_2): Linear(in_features=4096, out_features=1024, bias=True) (dropout): Dropout(p=0.1, inplace=False) (activation): ReLU() ) (norm1): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm2): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (norm3): LayerNorm((1024,), eps=1e-12, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) ) ) (criterion_att): LabelSmoothingLoss( (criterion): KLDivLoss() ) (ctc): CTC( (ctc_lo): Linear(in_features=1024, out_features=50002, bias=True) (ctc_loss): CTCLoss() ) ) Model summary: Class Name: ESPnetS2TModel Total Number of model parameters: 653.37 M Number of trainable parameters: 653.37 M (100.0%) Size: 2.61 GB Type: torch.float32 [cnode7-012:0/16] 2024-12-05 16:33:45,690 (abs_task:1387) INFO: Optimizer: Adadelta ( Parameter Group 0 capturable: False differentiable: False eps: 1e-06 foreach: None lr: 1.0 maximize: False rho: 0.9 weight_decay: 0 ) [cnode7-012:0/16] 2024-12-05 16:33:45,690 (abs_task:1388) INFO: Scheduler: None [cnode7-012:0/16] 2024-12-05 16:33:45,693 (abs_task:1397) INFO: Saving the configuration in exp_owsm/s2t_train_05b_ds_raw_bpe50000/config.yaml [cnode7-012:0/16] 2024-12-05 16:33:50,298 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') /mnt/home/williamchen/espnet/espnet2/train/deepspeed_trainer.py:240: InstrumentationWarning: @typechecked only supports instrumenting functions wrapped with @classmethod, @staticmethod or @property -- not typechecking espnet2.train.deepspeed_trainer.DeepSpeedTrainer.valid_one_epoch def valid_one_epoch( [2024-12-05 16:33:52,369] [INFO] [comm.py:652:init_distributed] cdb=None [2024-12-05 16:33:52,372] [INFO] [config.py:733:__init__] Config mesh_device None world_size = 16 /mnt/home/williamchen/espnet/espnet2/train/deepspeed_trainer.py:240: InstrumentationWarning: @typechecked only supports instrumenting functions wrapped with @classmethod, @staticmethod or @property -- not typechecking espnet2.train.deepspeed_trainer.DeepSpeedTrainer.valid_one_epoch def valid_one_epoch( [2024-12-05 16:33:52,645] [INFO] [comm.py:652:init_distributed] cdb=None [2024-12-05 16:33:52,646] [INFO] [config.py:733:__init__] Config mesh_device None world_size = 16 /mnt/home/williamchen/espnet/espnet2/train/deepspeed_trainer.py:240: InstrumentationWarning: @typechecked only supports instrumenting functions wrapped with @classmethod, @staticmethod or @property -- not typechecking espnet2.train.deepspeed_trainer.DeepSpeedTrainer.valid_one_epoch def valid_one_epoch( [2024-12-05 16:33:53,022] [INFO] [comm.py:652:init_distributed] cdb=None [2024-12-05 16:33:53,023] [INFO] [config.py:733:__init__] Config mesh_device None world_size = 16 /mnt/home/williamchen/espnet/espnet2/train/deepspeed_trainer.py:240: InstrumentationWarning: @typechecked only supports instrumenting functions wrapped with @classmethod, @staticmethod or @property -- not typechecking espnet2.train.deepspeed_trainer.DeepSpeedTrainer.valid_one_epoch def valid_one_epoch( [cnode7-012:0/16] 2024-12-05 16:33:53,259 (abs_task:1807) INFO: [valid] dataset: ESPnetDataset( speech: {"path": "dump/raw/dev_v3/wav.scp", "type": "kaldi_ark"} text_prev: {"path": "dump/raw/dev_v3/text.prev", "type": "text"} text_ctc: {"path": "dump/raw/dev_v3/text.ctc", "type": "text"} text: {"path": "dump/raw/dev_v3/text", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-05 16:33:53,259 (abs_task:1808) INFO: [valid] Batch sampler: SortedBatchSampler(N-batch=74743, batch_size=16, shape_file=exp_owsm/s2t_stats_raw_bpe50000/valid/speech_shape, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-05 16:33:53,265 (abs_task:1809) INFO: [valid] mini-batch sizes summary: N-batch=74743, mean=16.0, min=16, max=17 /mnt/home/williamchen/espnet/espnet2/train/deepspeed_trainer.py:240: InstrumentationWarning: @typechecked only supports instrumenting functions wrapped with @classmethod, @staticmethod or @property -- not typechecking espnet2.train.deepspeed_trainer.DeepSpeedTrainer.valid_one_epoch def valid_one_epoch( [2024-12-05 16:33:53,283] [INFO] [comm.py:652:init_distributed] cdb=None [2024-12-05 16:33:53,284] [INFO] [config.py:733:__init__] Config mesh_device None world_size = 16 /mnt/home/williamchen/espnet/espnet2/train/deepspeed_trainer.py:240: InstrumentationWarning: @typechecked only supports instrumenting functions wrapped with @classmethod, @staticmethod or @property -- not typechecking espnet2.train.deepspeed_trainer.DeepSpeedTrainer.valid_one_epoch def valid_one_epoch( [2024-12-05 16:33:53,291] [INFO] [comm.py:652:init_distributed] cdb=None [2024-12-05 16:33:53,292] [INFO] [config.py:733:__init__] Config mesh_device None world_size = 16 [2024-12-05 16:33:53,309] [INFO] [comm.py:652:init_distributed] cdb=None [2024-12-05 16:33:53,310] [INFO] [config.py:733:__init__] Config mesh_device None world_size = 16 /mnt/home/williamchen/espnet/espnet2/train/deepspeed_trainer.py:240: InstrumentationWarning: @typechecked only supports instrumenting functions wrapped with @classmethod, @staticmethod or @property -- not typechecking espnet2.train.deepspeed_trainer.DeepSpeedTrainer.valid_one_epoch def valid_one_epoch( [cnode7-012:0/16] 2024-12-05 16:33:53,321 (distributed_utils:129) WARNING: ================================================================= Found OMP_NUM_THREADS=1 in environment variables. With some advanced features, DeepSpeed may have heavy cpu workload so that OMP_NUM_THREADS=1 is not sufficient. Try to increase it in your path.sh ================================================================= [2024-12-05 16:33:53,321] [INFO] [comm.py:652:init_distributed] cdb=None [2024-12-05 16:33:53,322] [INFO] [logging.py:129:log_dist] [Rank 0] DeepSpeed info: version=0.15.3, git-hash=unknown, git-branch=unknown [2024-12-05 16:33:53,322] [INFO] [config.py:733:__init__] Config mesh_device None world_size = 16 cnode7-012:1272005:1272005 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to ^lo,docker,virbr,vmnet,vboxnet cnode7-012:1272005:1272005 [0] NCCL INFO Bootstrap : Using enp170s0np0:10.225.1.174<0> cnode7-012:1272005:1272005 [0] NCCL INFO NET/Plugin : dlerror=libnccl-net.so: cannot open shared object file: No such file or directory No plugin found (libnccl-net.so), using internal implementation cnode7-012:1272005:1272005 [0] NCCL INFO cudaDriverVersion 12020 NCCL version 2.20.5+cuda12.4 cnode7-012:1272006:1272006 [1] NCCL INFO cudaDriverVersion 12020 cnode7-012:1272006:1272006 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to ^lo,docker,virbr,vmnet,vboxnet cnode7-012:1272006:1272006 [1] NCCL INFO Bootstrap : Using enp170s0np0:10.225.1.174<0> cnode7-012:1272006:1272006 [1] NCCL INFO NET/Plugin : dlerror=libnccl-net.so: cannot open shared object file: No such file or directory No plugin found (libnccl-net.so), using internal implementation cnode7-012:1272008:1272008 [3] NCCL INFO cudaDriverVersion 12020 cnode7-012:1272008:1272008 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to ^lo,docker,virbr,vmnet,vboxnet cnode7-012:1272008:1272008 [3] NCCL INFO Bootstrap : Using enp170s0np0:10.225.1.174<0> cnode7-012:1272008:1272008 [3] NCCL INFO NET/Plugin : dlerror=libnccl-net.so: cannot open shared object file: No such file or directory No plugin found (libnccl-net.so), using internal implementation cnode7-012:1272007:1272007 [2] NCCL INFO cudaDriverVersion 12020 cnode7-012:1272007:1272007 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to ^lo,docker,virbr,vmnet,vboxnet cnode7-012:1272007:1272007 [2] NCCL INFO Bootstrap : Using enp170s0np0:10.225.1.174<0> cnode7-012:1272007:1272007 [2] NCCL INFO NET/Plugin : dlerror=libnccl-net.so: cannot open shared object file: No such file or directory No plugin found (libnccl-net.so), using internal implementation cnode7-012:1272012:1272012 [7] NCCL INFO cudaDriverVersion 12020 cnode7-012:1272012:1272012 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to ^lo,docker,virbr,vmnet,vboxnet cnode7-012:1272012:1272012 [7] NCCL INFO Bootstrap : Using enp170s0np0:10.225.1.174<0> cnode7-012:1272012:1272012 [7] NCCL INFO NET/Plugin : dlerror=libnccl-net.so: cannot open shared object file: No such file or directory No plugin found (libnccl-net.so), using internal implementation cnode7-012:1272009:1272009 [4] NCCL INFO cudaDriverVersion 12020 cnode7-012:1272009:1272009 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to ^lo,docker,virbr,vmnet,vboxnet cnode7-012:1272009:1272009 [4] NCCL INFO Bootstrap : Using enp170s0np0:10.225.1.174<0> cnode7-012:1272009:1272009 [4] NCCL INFO NET/Plugin : dlerror=libnccl-net.so: cannot open shared object file: No such file or directory No plugin found (libnccl-net.so), using internal implementation cnode7-012:1272011:1272011 [6] NCCL INFO cudaDriverVersion 12020 cnode7-012:1272011:1272011 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to ^lo,docker,virbr,vmnet,vboxnet cnode7-012:1272011:1272011 [6] NCCL INFO Bootstrap : Using enp170s0np0:10.225.1.174<0> cnode7-012:1272011:1272011 [6] NCCL INFO NET/Plugin : dlerror=libnccl-net.so: cannot open shared object file: No such file or directory No plugin found (libnccl-net.so), using internal implementation cnode7-012:1272007:1272925 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to ^lo,docker,virbr,vmnet,vboxnet cnode7-012:1272012:1272930 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to ^lo,docker,virbr,vmnet,vboxnet cnode7-012:1272009:1272927 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to ^lo,docker,virbr,vmnet,vboxnet cnode7-012:1272011:1272928 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to ^lo,docker,virbr,vmnet,vboxnet cnode7-012:1272006:1272926 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to ^lo,docker,virbr,vmnet,vboxnet cnode7-012:1272005:1272924 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to ^lo,docker,virbr,vmnet,vboxnet cnode7-012:1272008:1272929 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to ^lo,docker,virbr,vmnet,vboxnet cnode7-012:1272007:1272925 [2] NCCL INFO NET/IB : Using [0]mlx5_0:1/IB [1]mlx5_3:1/IB [2]mlx5_4:1/IB [3]mlx5_5:1/IB [4]mlx5_6:1/IB [5]mlx5_7:1/RoCE [6]mlx5_8:1/IB [7]mlx5_9:1/IB [8]mlx5_10:1/IB [RO]; OOB enp170s0np0:10.225.1.174<0> cnode7-012:1272007:1272925 [2] NCCL INFO Using non-device net plugin version 0 cnode7-012:1272007:1272925 [2] NCCL INFO Using network IB cnode7-012:1272012:1272930 [7] NCCL INFO NET/IB : Using [0]mlx5_0:1/IB [1]mlx5_3:1/IB [2]mlx5_4:1/IB [3]mlx5_5:1/IB [4]mlx5_6:1/IB [5]mlx5_7:1/RoCE [6]mlx5_8:1/IB [7]mlx5_9:1/IB [8]mlx5_10:1/IB [RO]; OOB enp170s0np0:10.225.1.174<0> cnode7-012:1272012:1272930 [7] NCCL INFO Using non-device net plugin version 0 cnode7-012:1272012:1272930 [7] NCCL INFO Using network IB cnode7-012:1272009:1272927 [4] NCCL INFO NET/IB : Using [0]mlx5_0:1/IB [1]mlx5_3:1/IB [2]mlx5_4:1/IB [3]mlx5_5:1/IB [4]mlx5_6:1/IB [5]mlx5_7:1/RoCE [6]mlx5_8:1/IB [7]mlx5_9:1/IB [8]mlx5_10:1/IB [RO]; OOB enp170s0np0:10.225.1.174<0> cnode7-012:1272009:1272927 [4] NCCL INFO Using non-device net plugin version 0 cnode7-012:1272009:1272927 [4] NCCL INFO Using network IB cnode7-012:1272011:1272928 [6] NCCL INFO NET/IB : Using [0]mlx5_0:1/IB [1]mlx5_3:1/IB [2]mlx5_4:1/IB [3]mlx5_5:1/IB [4]mlx5_6:1/IB [5]mlx5_7:1/RoCE [6]mlx5_8:1/IB [7]mlx5_9:1/IB [8]mlx5_10:1/IB [RO]; OOB enp170s0np0:10.225.1.174<0> cnode7-012:1272011:1272928 [6] NCCL INFO Using non-device net plugin version 0 cnode7-012:1272011:1272928 [6] NCCL INFO Using network IB cnode7-012:1272006:1272926 [1] NCCL INFO NET/IB : Using [0]mlx5_0:1/IB [1]mlx5_3:1/IB [2]mlx5_4:1/IB [3]mlx5_5:1/IB [4]mlx5_6:1/IB [5]mlx5_7:1/RoCE [6]mlx5_8:1/IB [7]mlx5_9:1/IB [8]mlx5_10:1/IB [RO]; OOB enp170s0np0:10.225.1.174<0> cnode7-012:1272006:1272926 [1] NCCL INFO Using non-device net plugin version 0 cnode7-012:1272006:1272926 [1] NCCL INFO Using network IB cnode7-012:1272005:1272924 [0] NCCL INFO NET/IB : Using [0]mlx5_0:1/IB [1]mlx5_3:1/IB [2]mlx5_4:1/IB [3]mlx5_5:1/IB [4]mlx5_6:1/IB [5]mlx5_7:1/RoCE [6]mlx5_8:1/IB [7]mlx5_9:1/IB [8]mlx5_10:1/IB [RO]; OOB enp170s0np0:10.225.1.174<0> cnode7-012:1272005:1272924 [0] NCCL INFO Using non-device net plugin version 0 cnode7-012:1272005:1272924 [0] NCCL INFO Using network IB cnode7-012:1272008:1272929 [3] NCCL INFO NET/IB : Using [0]mlx5_0:1/IB [1]mlx5_3:1/IB [2]mlx5_4:1/IB [3]mlx5_5:1/IB [4]mlx5_6:1/IB [5]mlx5_7:1/RoCE [6]mlx5_8:1/IB [7]mlx5_9:1/IB [8]mlx5_10:1/IB [RO]; OOB enp170s0np0:10.225.1.174<0> cnode7-012:1272008:1272929 [3] NCCL INFO Using non-device net plugin version 0 cnode7-012:1272008:1272929 [3] NCCL INFO Using network IB /mnt/home/williamchen/espnet/espnet2/train/deepspeed_trainer.py:240: InstrumentationWarning: @typechecked only supports instrumenting functions wrapped with @classmethod, @staticmethod or @property -- not typechecking espnet2.train.deepspeed_trainer.DeepSpeedTrainer.valid_one_epoch def valid_one_epoch( [2024-12-05 16:33:53,793] [INFO] [comm.py:652:init_distributed] cdb=None [2024-12-05 16:33:53,794] [INFO] [config.py:733:__init__] Config mesh_device None world_size = 16 cnode7-012:1272010:1272010 [5] NCCL INFO cudaDriverVersion 12020 cnode7-012:1272010:1272010 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to ^lo,docker,virbr,vmnet,vboxnet cnode7-012:1272010:1272010 [5] NCCL INFO Bootstrap : Using enp170s0np0:10.225.1.174<0> cnode7-012:1272010:1272010 [5] NCCL INFO NET/Plugin : dlerror=libnccl-net.so: cannot open shared object file: No such file or directory No plugin found (libnccl-net.so), using internal implementation /mnt/home/williamchen/espnet/espnet2/train/deepspeed_trainer.py:240: InstrumentationWarning: @typechecked only supports instrumenting functions wrapped with @classmethod, @staticmethod or @property -- not typechecking espnet2.train.deepspeed_trainer.DeepSpeedTrainer.valid_one_epoch def valid_one_epoch( /mnt/home/williamchen/espnet/espnet2/train/deepspeed_trainer.py:240: InstrumentationWarning: @typechecked only supports instrumenting functions wrapped with @classmethod, @staticmethod or @property -- not typechecking espnet2.train.deepspeed_trainer.DeepSpeedTrainer.valid_one_epoch def valid_one_epoch( /mnt/home/williamchen/espnet/espnet2/train/deepspeed_trainer.py:240: InstrumentationWarning: @typechecked only supports instrumenting functions wrapped with @classmethod, @staticmethod or @property -- not typechecking espnet2.train.deepspeed_trainer.DeepSpeedTrainer.valid_one_epoch def valid_one_epoch( [2024-12-05 16:33:55,054] [INFO] [comm.py:652:init_distributed] cdb=None [2024-12-05 16:33:55,057] [INFO] [config.py:733:__init__] Config mesh_device None world_size = 16 [2024-12-05 16:33:55,135] [INFO] [comm.py:652:init_distributed] cdb=None [2024-12-05 16:33:55,136] [INFO] [comm.py:652:init_distributed] cdb=None [2024-12-05 16:33:55,136] [INFO] [config.py:733:__init__] Config mesh_device None world_size = 16 [2024-12-05 16:33:55,137] [INFO] [config.py:733:__init__] Config mesh_device None world_size = 16 /mnt/home/williamchen/espnet/espnet2/train/deepspeed_trainer.py:240: InstrumentationWarning: @typechecked only supports instrumenting functions wrapped with @classmethod, @staticmethod or @property -- not typechecking espnet2.train.deepspeed_trainer.DeepSpeedTrainer.valid_one_epoch def valid_one_epoch( /mnt/home/williamchen/espnet/espnet2/train/deepspeed_trainer.py:240: InstrumentationWarning: @typechecked only supports instrumenting functions wrapped with @classmethod, @staticmethod or @property -- not typechecking espnet2.train.deepspeed_trainer.DeepSpeedTrainer.valid_one_epoch def valid_one_epoch( cnode7-012:1272010:1272998 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to ^lo,docker,virbr,vmnet,vboxnet [2024-12-05 16:33:55,209] [INFO] [comm.py:652:init_distributed] cdb=None [2024-12-05 16:33:55,210] [INFO] [config.py:733:__init__] Config mesh_device None world_size = 16 cnode7-013:736488:736488 [7] NCCL INFO cudaDriverVersion 12020 cnode7-013:736488:736488 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to ^lo,docker,virbr,vmnet,vboxnet cnode7-013:736488:736488 [7] NCCL INFO Bootstrap : Using enp170s0np0:10.225.1.175<0> cnode7-013:736488:736488 [7] NCCL INFO NET/Plugin : dlerror=libnccl-net.so: cannot open shared object file: No such file or directory No plugin found (libnccl-net.so), using internal implementation cnode7-013:736487:736487 [6] NCCL INFO cudaDriverVersion 12020 cnode7-013:736487:736487 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to ^lo,docker,virbr,vmnet,vboxnet cnode7-013:736487:736487 [6] NCCL INFO Bootstrap : Using enp170s0np0:10.225.1.175<0> cnode7-013:736487:736487 [6] NCCL INFO NET/Plugin : dlerror=libnccl-net.so: cannot open shared object file: No such file or directory No plugin found (libnccl-net.so), using internal implementation cnode7-013:736485:736485 [4] NCCL INFO cudaDriverVersion 12020 cnode7-013:736485:736485 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to ^lo,docker,virbr,vmnet,vboxnet cnode7-013:736485:736485 [4] NCCL INFO Bootstrap : Using enp170s0np0:10.225.1.175<0> cnode7-013:736485:736485 [4] NCCL INFO NET/Plugin : dlerror=libnccl-net.so: cannot open shared object file: No such file or directory No plugin found (libnccl-net.so), using internal implementation cnode7-013:736481:736481 [0] NCCL INFO cudaDriverVersion 12020 cnode7-013:736481:736481 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to ^lo,docker,virbr,vmnet,vboxnet cnode7-013:736481:736481 [0] NCCL INFO Bootstrap : Using enp170s0np0:10.225.1.175<0> cnode7-013:736481:736481 [0] NCCL INFO NET/Plugin : dlerror=libnccl-net.so: cannot open shared object file: No such file or directory No plugin found (libnccl-net.so), using internal implementation cnode7-012:1272010:1272998 [5] NCCL INFO NET/IB : Using [0]mlx5_0:1/IB [1]mlx5_3:1/IB [2]mlx5_4:1/IB [3]mlx5_5:1/IB [4]mlx5_6:1/IB [5]mlx5_7:1/RoCE [6]mlx5_8:1/IB [7]mlx5_9:1/IB [8]mlx5_10:1/IB [RO]; OOB enp170s0np0:10.225.1.174<0> cnode7-012:1272010:1272998 [5] NCCL INFO Using non-device net plugin version 0 cnode7-012:1272010:1272998 [5] NCCL INFO Using network IB cnode7-013:736485:737373 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to ^lo,docker,virbr,vmnet,vboxnet [2024-12-05 16:33:55,380] [INFO] [comm.py:652:init_distributed] cdb=None [2024-12-05 16:33:55,382] [INFO] [config.py:733:__init__] Config mesh_device None world_size = 16 /mnt/home/williamchen/espnet/espnet2/train/deepspeed_trainer.py:240: InstrumentationWarning: @typechecked only supports instrumenting functions wrapped with @classmethod, @staticmethod or @property -- not typechecking espnet2.train.deepspeed_trainer.DeepSpeedTrainer.valid_one_epoch def valid_one_epoch( cnode7-013:736488:737372 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to ^lo,docker,virbr,vmnet,vboxnet cnode7-013:736487:737367 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to ^lo,docker,virbr,vmnet,vboxnet cnode7-013:736485:737373 [4] NCCL INFO NET/IB : Using [0]mlx5_0:1/IB [1]mlx5_3:1/IB [2]mlx5_4:1/IB [3]mlx5_5:1/IB [4]mlx5_6:1/IB [5]mlx5_7:1/RoCE [6]mlx5_8:1/IB [7]mlx5_9:1/IB [8]mlx5_10:1/IB [RO]; OOB enp170s0np0:10.225.1.175<0> cnode7-013:736485:737373 [4] NCCL INFO Using non-device net plugin version 0 cnode7-013:736485:737373 [4] NCCL INFO Using network IB cnode7-013:736481:737376 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to ^lo,docker,virbr,vmnet,vboxnet cnode7-013:736484:736484 [3] NCCL INFO cudaDriverVersion 12020 cnode7-013:736484:736484 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to ^lo,docker,virbr,vmnet,vboxnet cnode7-013:736484:736484 [3] NCCL INFO Bootstrap : Using enp170s0np0:10.225.1.175<0> cnode7-013:736484:736484 [3] NCCL INFO NET/Plugin : dlerror=libnccl-net.so: cannot open shared object file: No such file or directory No plugin found (libnccl-net.so), using internal implementation cnode7-013:736488:737372 [7] NCCL INFO NET/IB : Using [0]mlx5_0:1/IB [1]mlx5_3:1/IB [2]mlx5_4:1/IB [3]mlx5_5:1/IB [4]mlx5_6:1/IB [5]mlx5_7:1/RoCE [6]mlx5_8:1/IB [7]mlx5_9:1/IB [8]mlx5_10:1/IB [RO]; OOB enp170s0np0:10.225.1.175<0> cnode7-013:736488:737372 [7] NCCL INFO Using non-device net plugin version 0 cnode7-013:736488:737372 [7] NCCL INFO Using network IB cnode7-013:736487:737367 [6] NCCL INFO NET/IB : Using [0]mlx5_0:1/IB [1]mlx5_3:1/IB [2]mlx5_4:1/IB [3]mlx5_5:1/IB [4]mlx5_6:1/IB [5]mlx5_7:1/RoCE [6]mlx5_8:1/IB [7]mlx5_9:1/IB [8]mlx5_10:1/IB [RO]; OOB enp170s0np0:10.225.1.175<0> cnode7-013:736487:737367 [6] NCCL INFO Using non-device net plugin version 0 cnode7-013:736487:737367 [6] NCCL INFO Using network IB /mnt/home/williamchen/espnet/espnet2/train/deepspeed_trainer.py:240: InstrumentationWarning: @typechecked only supports instrumenting functions wrapped with @classmethod, @staticmethod or @property -- not typechecking espnet2.train.deepspeed_trainer.DeepSpeedTrainer.valid_one_epoch def valid_one_epoch( [2024-12-05 16:33:55,569] [INFO] [comm.py:652:init_distributed] cdb=None [2024-12-05 16:33:55,570] [INFO] [config.py:733:__init__] Config mesh_device None world_size = 16 cnode7-013:736481:737376 [0] NCCL INFO NET/IB : Using [0]mlx5_0:1/IB [1]mlx5_3:1/IB [2]mlx5_4:1/IB [3]mlx5_5:1/IB [4]mlx5_6:1/IB [5]mlx5_7:1/RoCE [6]mlx5_8:1/IB [7]mlx5_9:1/IB [8]mlx5_10:1/IB [RO]; OOB enp170s0np0:10.225.1.175<0> cnode7-013:736481:737376 [0] NCCL INFO Using non-device net plugin version 0 cnode7-013:736481:737376 [0] NCCL INFO Using network IB [2024-12-05 16:33:55,608] [INFO] [comm.py:652:init_distributed] cdb=None [2024-12-05 16:33:55,609] [INFO] [config.py:733:__init__] Config mesh_device None world_size = 16 /mnt/home/williamchen/espnet/espnet2/train/deepspeed_trainer.py:240: InstrumentationWarning: @typechecked only supports instrumenting functions wrapped with @classmethod, @staticmethod or @property -- not typechecking espnet2.train.deepspeed_trainer.DeepSpeedTrainer.valid_one_epoch def valid_one_epoch( cnode7-013:736483:736483 [2] NCCL INFO cudaDriverVersion 12020 cnode7-013:736483:736483 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to ^lo,docker,virbr,vmnet,vboxnet cnode7-013:736483:736483 [2] NCCL INFO Bootstrap : Using enp170s0np0:10.225.1.175<0> cnode7-013:736483:736483 [2] NCCL INFO NET/Plugin : dlerror=libnccl-net.so: cannot open shared object file: No such file or directory No plugin found (libnccl-net.so), using internal implementation cnode7-013:736484:737409 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to ^lo,docker,virbr,vmnet,vboxnet cnode7-013:736486:736486 [5] NCCL INFO cudaDriverVersion 12020 cnode7-013:736486:736486 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to ^lo,docker,virbr,vmnet,vboxnet cnode7-013:736486:736486 [5] NCCL INFO Bootstrap : Using enp170s0np0:10.225.1.175<0> cnode7-013:736486:736486 [5] NCCL INFO NET/Plugin : dlerror=libnccl-net.so: cannot open shared object file: No such file or directory No plugin found (libnccl-net.so), using internal implementation [2024-12-05 16:33:56,271] [INFO] [comm.py:652:init_distributed] cdb=None [2024-12-05 16:33:56,273] [INFO] [config.py:733:__init__] Config mesh_device None world_size = 16 cnode7-013:736484:737409 [3] NCCL INFO NET/IB : Using [0]mlx5_0:1/IB [1]mlx5_3:1/IB [2]mlx5_4:1/IB [3]mlx5_5:1/IB [4]mlx5_6:1/IB [5]mlx5_7:1/RoCE [6]mlx5_8:1/IB [7]mlx5_9:1/IB [8]mlx5_10:1/IB [RO]; OOB enp170s0np0:10.225.1.175<0> cnode7-013:736484:737409 [3] NCCL INFO Using non-device net plugin version 0 cnode7-013:736484:737409 [3] NCCL INFO Using network IB cnode7-013:736482:736482 [1] NCCL INFO cudaDriverVersion 12020 cnode7-013:736482:736482 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to ^lo,docker,virbr,vmnet,vboxnet cnode7-013:736482:736482 [1] NCCL INFO Bootstrap : Using enp170s0np0:10.225.1.175<0> cnode7-013:736482:736482 [1] NCCL INFO NET/Plugin : dlerror=libnccl-net.so: cannot open shared object file: No such file or directory No plugin found (libnccl-net.so), using internal implementation cnode7-013:736486:737423 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to ^lo,docker,virbr,vmnet,vboxnet cnode7-013:736483:737422 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to ^lo,docker,virbr,vmnet,vboxnet cnode7-013:736486:737423 [5] NCCL INFO NET/IB : Using [0]mlx5_0:1/IB [1]mlx5_3:1/IB [2]mlx5_4:1/IB [3]mlx5_5:1/IB [4]mlx5_6:1/IB [5]mlx5_7:1/RoCE [6]mlx5_8:1/IB [7]mlx5_9:1/IB [8]mlx5_10:1/IB [RO]; OOB enp170s0np0:10.225.1.175<0> cnode7-013:736486:737423 [5] NCCL INFO Using non-device net plugin version 0 cnode7-013:736486:737423 [5] NCCL INFO Using network IB cnode7-013:736483:737422 [2] NCCL INFO NET/IB : Using [0]mlx5_0:1/IB [1]mlx5_3:1/IB [2]mlx5_4:1/IB [3]mlx5_5:1/IB [4]mlx5_6:1/IB [5]mlx5_7:1/RoCE [6]mlx5_8:1/IB [7]mlx5_9:1/IB [8]mlx5_10:1/IB [RO]; OOB enp170s0np0:10.225.1.175<0> cnode7-013:736483:737422 [2] NCCL INFO Using non-device net plugin version 0 cnode7-013:736483:737422 [2] NCCL INFO Using network IB cnode7-013:736482:737436 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to ^lo,docker,virbr,vmnet,vboxnet cnode7-013:736482:737436 [1] NCCL INFO NET/IB : Using [0]mlx5_0:1/IB [1]mlx5_3:1/IB [2]mlx5_4:1/IB [3]mlx5_5:1/IB [4]mlx5_6:1/IB [5]mlx5_7:1/RoCE [6]mlx5_8:1/IB [7]mlx5_9:1/IB [8]mlx5_10:1/IB [RO]; OOB enp170s0np0:10.225.1.175<0> cnode7-013:736482:737436 [1] NCCL INFO Using non-device net plugin version 0 cnode7-013:736482:737436 [1] NCCL INFO Using network IB cnode7-012:1272006:1272926 [1] NCCL INFO comm 0x55558161ded0 rank 1 nranks 16 cudaDev 1 nvmlDev 1 busId 43000 commId 0x4e42d6174620c76 - Init START cnode7-012:1272005:1272924 [0] NCCL INFO comm 0x555563bd8350 rank 0 nranks 16 cudaDev 0 nvmlDev 0 busId 1b000 commId 0x4e42d6174620c76 - Init START cnode7-012:1272010:1272998 [5] NCCL INFO comm 0x555572620330 rank 5 nranks 16 cudaDev 5 nvmlDev 5 busId c3000 commId 0x4e42d6174620c76 - Init START cnode7-013:736484:737409 [3] NCCL INFO comm 0x55557b419c80 rank 11 nranks 16 cudaDev 3 nvmlDev 3 busId 61000 commId 0x4e42d6174620c76 - Init START cnode7-013:736488:737372 [7] NCCL INFO comm 0x555560c15d00 rank 15 nranks 16 cudaDev 7 nvmlDev 7 busId df000 commId 0x4e42d6174620c76 - Init START cnode7-013:736487:737367 [6] NCCL INFO comm 0x555561a1a9e0 rank 14 nranks 16 cudaDev 6 nvmlDev 6 busId d1000 commId 0x4e42d6174620c76 - Init START cnode7-012:1272008:1272929 [3] NCCL INFO comm 0x55557b81cee0 rank 3 nranks 16 cudaDev 3 nvmlDev 3 busId 61000 commId 0x4e42d6174620c76 - Init START cnode7-013:736482:737436 [1] NCCL INFO comm 0x555580e21bb0 rank 9 nranks 16 cudaDev 1 nvmlDev 1 busId 43000 commId 0x4e42d6174620c76 - Init START cnode7-012:1272007:1272925 [2] NCCL INFO comm 0x55558061a960 rank 2 nranks 16 cudaDev 2 nvmlDev 2 busId 52000 commId 0x4e42d6174620c76 - Init START cnode7-013:736485:737373 [4] NCCL INFO comm 0x555560c19b70 rank 12 nranks 16 cudaDev 4 nvmlDev 4 busId 9d000 commId 0x4e42d6174620c76 - Init START cnode7-013:736486:737423 [5] NCCL INFO comm 0x555581618ce0 rank 13 nranks 16 cudaDev 5 nvmlDev 5 busId c3000 commId 0x4e42d6174620c76 - Init START cnode7-012:1272012:1272930 [7] NCCL INFO comm 0x55557b820a90 rank 7 nranks 16 cudaDev 7 nvmlDev 7 busId df000 commId 0x4e42d6174620c76 - Init START cnode7-013:736481:737376 [0] NCCL INFO comm 0x555563b6a610 rank 8 nranks 16 cudaDev 0 nvmlDev 0 busId 1b000 commId 0x4e42d6174620c76 - Init START cnode7-012:1272009:1272927 [4] NCCL INFO comm 0x55558161a360 rank 4 nranks 16 cudaDev 4 nvmlDev 4 busId 9d000 commId 0x4e42d6174620c76 - Init START cnode7-012:1272011:1272928 [6] NCCL INFO comm 0x55557c81dd70 rank 6 nranks 16 cudaDev 6 nvmlDev 6 busId d1000 commId 0x4e42d6174620c76 - Init START cnode7-013:736483:737422 [2] NCCL INFO comm 0x55558161e2b0 rank 10 nranks 16 cudaDev 2 nvmlDev 2 busId 52000 commId 0x4e42d6174620c76 - Init START cnode7-013:736488:737372 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffffff00,00000000,0000ffff,ffffffff,ff000000,00000000 cnode7-013:736488:737372 [7] NCCL INFO NVLS multicast support is available on dev 7 cnode7-013:736483:737422 [2] NCCL INFO Setting affinity for GPU 2 to ff,ffffffff,ffff0000,00000000,00ffffff,ffffffff cnode7-013:736483:737422 [2] NCCL INFO NVLS multicast support is available on dev 2 cnode7-013:736487:737367 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffffff00,00000000,0000ffff,ffffffff,ff000000,00000000 cnode7-013:736487:737367 [6] NCCL INFO NVLS multicast support is available on dev 6 cnode7-013:736482:737436 [1] NCCL INFO Setting affinity for GPU 1 to ff,ffffffff,ffff0000,00000000,00ffffff,ffffffff cnode7-013:736482:737436 [1] NCCL INFO NVLS multicast support is available on dev 1 cnode7-012:1272007:1272925 [2] NCCL INFO Setting affinity for GPU 2 to ff,ffffffff,ffff0000,00000000,00ffffff,ffffffff cnode7-012:1272007:1272925 [2] NCCL INFO NVLS multicast support is available on dev 2 cnode7-013:736485:737373 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffffff00,00000000,0000ffff,ffffffff,ff000000,00000000 cnode7-013:736485:737373 [4] NCCL INFO NVLS multicast support is available on dev 4 cnode7-012:1272012:1272930 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffffff00,00000000,0000ffff,ffffffff,ff000000,00000000 cnode7-012:1272012:1272930 [7] NCCL INFO NVLS multicast support is available on dev 7 cnode7-013:736481:737376 [0] NCCL INFO Setting affinity for GPU 0 to ff,ffffffff,ffff0000,00000000,00ffffff,ffffffff cnode7-013:736481:737376 [0] NCCL INFO NVLS multicast support is available on dev 0 cnode7-012:1272009:1272927 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffffff00,00000000,0000ffff,ffffffff,ff000000,00000000 cnode7-012:1272009:1272927 [4] NCCL INFO NVLS multicast support is available on dev 4 cnode7-012:1272011:1272928 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffffff00,00000000,0000ffff,ffffffff,ff000000,00000000 cnode7-012:1272011:1272928 [6] NCCL INFO NVLS multicast support is available on dev 6 cnode7-013:736486:737423 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffffff00,00000000,0000ffff,ffffffff,ff000000,00000000 cnode7-013:736486:737423 [5] NCCL INFO NVLS multicast support is available on dev 5 cnode7-013:736486:737423 [5] NCCL INFO comm 0x555581618ce0 rank 13 nRanks 16 nNodes 2 localRanks 8 localRank 5 MNNVL 0 cnode7-013:736486:737423 [5] NCCL INFO Trees [0] 14/-1/-1->13->12 [1] 14/-1/-1->13->12 [2] 14/-1/-1->13->12 [3] 14/-1/-1->13->12 [4] 14/-1/-1->13->12 [5] 14/-1/-1->13->5 [6] -1/-1/-1->13->12 [7] 14/-1/-1->13->12 [8] 14/-1/-1->13->12 [9] 14/-1/-1->13->12 [10] 14/-1/-1->13->12 [11] 14/-1/-1->13->12 [12] 14/-1/-1->13->12 [13] 14/5/-1->13->-1 [14] -1/-1/-1->13->12 [15] 14/-1/-1->13->12 cnode7-013:736486:737423 [5] NCCL INFO P2P Chunksize set to 131072 cnode7-013:736487:737367 [6] NCCL INFO comm 0x555561a1a9e0 rank 14 nRanks 16 nNodes 2 localRanks 8 localRank 6 MNNVL 0 cnode7-013:736487:737367 [6] NCCL INFO Trees [0] 15/-1/-1->14->13 [1] 15/-1/-1->14->13 [2] 15/-1/-1->14->13 [3] 15/-1/-1->14->13 [4] 15/-1/-1->14->13 [5] 15/-1/-1->14->13 [6] 15/-1/-1->14->6 [7] -1/-1/-1->14->13 [8] 15/-1/-1->14->13 [9] 15/-1/-1->14->13 [10] 15/-1/-1->14->13 [11] 15/-1/-1->14->13 [12] 15/-1/-1->14->13 [13] 15/-1/-1->14->13 [14] 15/6/-1->14->-1 [15] -1/-1/-1->14->13 cnode7-013:736487:737367 [6] NCCL INFO P2P Chunksize set to 131072 cnode7-012:1272006:1272926 [1] NCCL INFO Setting affinity for GPU 1 to ff,ffffffff,ffff0000,00000000,00ffffff,ffffffff cnode7-012:1272006:1272926 [1] NCCL INFO NVLS multicast support is available on dev 1 cnode7-012:1272006:1272926 [1] NCCL INFO comm 0x55558161ded0 rank 1 nRanks 16 nNodes 2 localRanks 8 localRank 1 MNNVL 0 cnode7-012:1272006:1272926 [1] NCCL INFO NVLS Head 0: 0 8 cnode7-012:1272006:1272926 [1] NCCL INFO NVLS Head 1: 1 9 cnode7-012:1272006:1272926 [1] NCCL INFO NVLS Head 2: 2 10 cnode7-012:1272006:1272926 [1] NCCL INFO NVLS Head 3: 3 11 cnode7-012:1272006:1272926 [1] NCCL INFO NVLS Head 4: 4 12 cnode7-012:1272006:1272926 [1] NCCL INFO NVLS Head 5: 5 13 cnode7-012:1272006:1272926 [1] NCCL INFO NVLS Head 6: 6 14 cnode7-012:1272006:1272926 [1] NCCL INFO NVLS Head 7: 7 15 cnode7-012:1272006:1272926 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/9/-1->1->-1 [2] -1/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->9 [10] -1/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 cnode7-012:1272006:1272926 [1] NCCL INFO P2P Chunksize set to 131072 cnode7-013:736488:737372 [7] NCCL INFO comm 0x555560c15d00 rank 15 nRanks 16 nNodes 2 localRanks 8 localRank 7 MNNVL 0 cnode7-013:736488:737372 [7] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] 8/-1/-1->15->14 [2] 8/-1/-1->15->14 [3] 8/-1/-1->15->14 [4] 8/-1/-1->15->14 [5] 8/-1/-1->15->14 [6] 8/-1/-1->15->14 [7] 8/-1/-1->15->7 [8] -1/-1/-1->15->14 [9] 8/-1/-1->15->14 [10] 8/-1/-1->15->14 [11] 8/-1/-1->15->14 [12] 8/-1/-1->15->14 [13] 8/-1/-1->15->14 [14] 8/-1/-1->15->14 [15] 8/7/-1->15->-1 cnode7-013:736488:737372 [7] NCCL INFO P2P Chunksize set to 131072 cnode7-013:736481:737376 [0] NCCL INFO comm 0x555563b6a610 rank 8 nRanks 16 nNodes 2 localRanks 8 localRank 0 MNNVL 0 cnode7-013:736481:737376 [0] NCCL INFO Trees [0] 9/-1/-1->8->0 [1] -1/-1/-1->8->15 [2] 9/-1/-1->8->15 [3] 9/-1/-1->8->15 [4] 9/-1/-1->8->15 [5] 9/-1/-1->8->15 [6] 9/-1/-1->8->15 [7] 9/-1/-1->8->15 [8] 9/0/-1->8->-1 [9] -1/-1/-1->8->15 [10] 9/-1/-1->8->15 [11] 9/-1/-1->8->15 [12] 9/-1/-1->8->15 [13] 9/-1/-1->8->15 [14] 9/-1/-1->8->15 [15] 9/-1/-1->8->15 cnode7-013:736481:737376 [0] NCCL INFO P2P Chunksize set to 131072 cnode7-013:736485:737373 [4] NCCL INFO comm 0x555560c19b70 rank 12 nRanks 16 nNodes 2 localRanks 8 localRank 4 MNNVL 0 cnode7-013:736485:737373 [4] NCCL INFO Trees [0] 13/-1/-1->12->11 [1] 13/-1/-1->12->11 [2] 13/-1/-1->12->11 [3] 13/-1/-1->12->11 [4] 13/-1/-1->12->4 [5] -1/-1/-1->12->11 [6] 13/-1/-1->12->11 [7] 13/-1/-1->12->11 [8] 13/-1/-1->12->11 [9] 13/-1/-1->12->11 [10] 13/-1/-1->12->11 [11] 13/-1/-1->12->11 [12] 13/4/-1->12->-1 [13] -1/-1/-1->12->11 [14] 13/-1/-1->12->11 [15] 13/-1/-1->12->11 cnode7-013:736485:737373 [4] NCCL INFO P2P Chunksize set to 131072 cnode7-012:1272005:1272924 [0] NCCL INFO Setting affinity for GPU 0 to ff,ffffffff,ffff0000,00000000,00ffffff,ffffffff cnode7-012:1272005:1272924 [0] NCCL INFO NVLS multicast support is available on dev 0 cnode7-012:1272005:1272924 [0] NCCL INFO comm 0x555563bd8350 rank 0 nRanks 16 nNodes 2 localRanks 8 localRank 0 MNNVL 0 cnode7-012:1272005:1272924 [0] NCCL INFO NVLS Head 0: 0 8 cnode7-012:1272005:1272924 [0] NCCL INFO NVLS Head 1: 1 9 cnode7-012:1272005:1272924 [0] NCCL INFO NVLS Head 2: 2 10 cnode7-012:1272005:1272924 [0] NCCL INFO NVLS Head 3: 3 11 cnode7-012:1272005:1272924 [0] NCCL INFO NVLS Head 4: 4 12 cnode7-012:1272005:1272924 [0] NCCL INFO NVLS Head 5: 5 13 cnode7-012:1272005:1272924 [0] NCCL INFO NVLS Head 6: 6 14 cnode7-012:1272005:1272924 [0] NCCL INFO NVLS Head 7: 7 15 cnode7-012:1272005:1272924 [0] NCCL INFO Channel 00/16 : 0 7 6 5 4 3 2 1 9 10 11 12 13 14 15 8 cnode7-012:1272005:1272924 [0] NCCL INFO Channel 01/16 : 0 8 15 14 13 12 11 10 9 1 2 3 4 5 6 7 cnode7-012:1272005:1272924 [0] NCCL INFO Channel 02/16 : 0 7 6 5 4 3 11 12 13 14 15 8 9 10 2 1 cnode7-012:1272005:1272924 [0] NCCL INFO Channel 03/16 : 0 1 2 10 9 8 15 14 13 12 11 3 4 5 6 7 cnode7-012:1272005:1272924 [0] NCCL INFO Channel 04/16 : 0 7 6 5 13 14 15 8 9 10 11 12 4 3 2 1 cnode7-012:1272005:1272924 [0] NCCL INFO Channel 05/16 : 0 1 2 3 4 12 11 10 9 8 15 14 13 5 6 7 cnode7-012:1272005:1272924 [0] NCCL INFO Channel 06/16 : 0 7 15 8 9 10 11 12 13 14 6 5 4 3 2 1 cnode7-012:1272005:1272924 [0] NCCL INFO Channel 07/16 : 0 1 2 3 4 5 6 14 13 12 11 10 9 8 15 7 cnode7-012:1272005:1272924 [0] NCCL INFO Channel 08/16 : 0 7 6 5 4 3 2 1 9 10 11 12 13 14 15 8 cnode7-012:1272005:1272924 [0] NCCL INFO Channel 09/16 : 0 8 15 14 13 12 11 10 9 1 2 3 4 5 6 7 cnode7-012:1272005:1272924 [0] NCCL INFO Channel 10/16 : 0 7 6 5 4 3 11 12 13 14 15 8 9 10 2 1 cnode7-012:1272005:1272924 [0] NCCL INFO Channel 11/16 : 0 1 2 10 9 8 15 14 13 12 11 3 4 5 6 7 cnode7-012:1272005:1272924 [0] NCCL INFO Channel 12/16 : 0 7 6 5 13 14 15 8 9 10 11 12 4 3 2 1 cnode7-012:1272005:1272924 [0] NCCL INFO Channel 13/16 : 0 1 2 3 4 12 11 10 9 8 15 14 13 5 6 7 cnode7-012:1272005:1272924 [0] NCCL INFO Channel 14/16 : 0 7 15 8 9 10 11 12 13 14 6 5 4 3 2 1 cnode7-012:1272005:1272924 [0] NCCL INFO Channel 15/16 : 0 1 2 3 4 5 6 14 13 12 11 10 9 8 15 7 cnode7-012:1272005:1272924 [0] NCCL INFO Trees [0] 1/8/-1->0->-1 [1] -1/-1/-1->0->7 [2] 1/-1/-1->0->7 [3] 1/-1/-1->0->7 [4] 1/-1/-1->0->7 [5] 1/-1/-1->0->7 [6] 1/-1/-1->0->7 [7] 1/-1/-1->0->7 [8] 1/-1/-1->0->8 [9] -1/-1/-1->0->7 [10] 1/-1/-1->0->7 [11] 1/-1/-1->0->7 [12] 1/-1/-1->0->7 [13] 1/-1/-1->0->7 [14] 1/-1/-1->0->7 [15] 1/-1/-1->0->7 cnode7-012:1272005:1272924 [0] NCCL INFO P2P Chunksize set to 131072 cnode7-013:736482:737436 [1] NCCL INFO comm 0x555580e21bb0 rank 9 nRanks 16 nNodes 2 localRanks 8 localRank 1 MNNVL 0 cnode7-013:736482:737436 [1] NCCL INFO Trees [0] 10/-1/-1->9->8 [1] 10/-1/-1->9->1 [2] -1/-1/-1->9->8 [3] 10/-1/-1->9->8 [4] 10/-1/-1->9->8 [5] 10/-1/-1->9->8 [6] 10/-1/-1->9->8 [7] 10/-1/-1->9->8 [8] 10/-1/-1->9->8 [9] 10/1/-1->9->-1 [10] -1/-1/-1->9->8 [11] 10/-1/-1->9->8 [12] 10/-1/-1->9->8 [13] 10/-1/-1->9->8 [14] 10/-1/-1->9->8 [15] 10/-1/-1->9->8 cnode7-013:736482:737436 [1] NCCL INFO P2P Chunksize set to 131072 cnode7-012:1272010:1272998 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffffff00,00000000,0000ffff,ffffffff,ff000000,00000000 cnode7-012:1272010:1272998 [5] NCCL INFO NVLS multicast support is available on dev 5 cnode7-012:1272010:1272998 [5] NCCL INFO comm 0x555572620330 rank 5 nRanks 16 nNodes 2 localRanks 8 localRank 5 MNNVL 0 cnode7-012:1272010:1272998 [5] NCCL INFO NVLS Head 0: 0 8 cnode7-012:1272010:1272998 [5] NCCL INFO NVLS Head 1: 1 9 cnode7-012:1272010:1272998 [5] NCCL INFO NVLS Head 2: 2 10 cnode7-012:1272010:1272998 [5] NCCL INFO NVLS Head 3: 3 11 cnode7-012:1272010:1272998 [5] NCCL INFO NVLS Head 4: 4 12 cnode7-012:1272010:1272998 [5] NCCL INFO NVLS Head 5: 5 13 cnode7-012:1272010:1272998 [5] NCCL INFO NVLS Head 6: 6 14 cnode7-012:1272010:1272998 [5] NCCL INFO NVLS Head 7: 7 15 cnode7-012:1272010:1272998 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/13/-1->5->-1 [6] -1/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->13 [14] -1/-1/-1->5->4 [15] 6/-1/-1->5->4 cnode7-012:1272010:1272998 [5] NCCL INFO P2P Chunksize set to 131072 cnode7-013:736484:737409 [3] NCCL INFO Setting affinity for GPU 3 to ff,ffffffff,ffff0000,00000000,00ffffff,ffffffff cnode7-013:736484:737409 [3] NCCL INFO NVLS multicast support is available on dev 3 cnode7-013:736484:737409 [3] NCCL INFO comm 0x55557b419c80 rank 11 nRanks 16 nNodes 2 localRanks 8 localRank 3 MNNVL 0 cnode7-013:736484:737409 [3] NCCL INFO Trees [0] 12/-1/-1->11->10 [1] 12/-1/-1->11->10 [2] 12/-1/-1->11->10 [3] 12/-1/-1->11->3 [4] -1/-1/-1->11->10 [5] 12/-1/-1->11->10 [6] 12/-1/-1->11->10 [7] 12/-1/-1->11->10 [8] 12/-1/-1->11->10 [9] 12/-1/-1->11->10 [10] 12/-1/-1->11->10 [11] 12/3/-1->11->-1 [12] -1/-1/-1->11->10 [13] 12/-1/-1->11->10 [14] 12/-1/-1->11->10 [15] 12/-1/-1->11->10 cnode7-013:736484:737409 [3] NCCL INFO P2P Chunksize set to 131072 cnode7-012:1272008:1272929 [3] NCCL INFO Setting affinity for GPU 3 to ff,ffffffff,ffff0000,00000000,00ffffff,ffffffff cnode7-012:1272008:1272929 [3] NCCL INFO NVLS multicast support is available on dev 3 cnode7-012:1272008:1272929 [3] NCCL INFO comm 0x55557b81cee0 rank 3 nRanks 16 nNodes 2 localRanks 8 localRank 3 MNNVL 0 cnode7-012:1272008:1272929 [3] NCCL INFO NVLS Head 0: 0 8 cnode7-012:1272008:1272929 [3] NCCL INFO NVLS Head 1: 1 9 cnode7-012:1272008:1272929 [3] NCCL INFO NVLS Head 2: 2 10 cnode7-012:1272008:1272929 [3] NCCL INFO NVLS Head 3: 3 11 cnode7-012:1272008:1272929 [3] NCCL INFO NVLS Head 4: 4 12 cnode7-012:1272008:1272929 [3] NCCL INFO NVLS Head 5: 5 13 cnode7-012:1272008:1272929 [3] NCCL INFO NVLS Head 6: 6 14 cnode7-012:1272008:1272929 [3] NCCL INFO NVLS Head 7: 7 15 cnode7-012:1272008:1272929 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/11/-1->3->-1 [4] -1/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->11 [12] -1/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 cnode7-012:1272008:1272929 [3] NCCL INFO P2P Chunksize set to 131072 cnode7-012:1272012:1272930 [7] NCCL INFO comm 0x55557b820a90 rank 7 nRanks 16 nNodes 2 localRanks 8 localRank 7 MNNVL 0 cnode7-012:1272012:1272930 [7] NCCL INFO NVLS Head 0: 0 8 cnode7-012:1272012:1272930 [7] NCCL INFO NVLS Head 1: 1 9 cnode7-012:1272012:1272930 [7] NCCL INFO NVLS Head 2: 2 10 cnode7-012:1272012:1272930 [7] NCCL INFO NVLS Head 3: 3 11 cnode7-012:1272012:1272930 [7] NCCL INFO NVLS Head 4: 4 12 cnode7-012:1272012:1272930 [7] NCCL INFO NVLS Head 5: 5 13 cnode7-012:1272012:1272930 [7] NCCL INFO NVLS Head 6: 6 14 cnode7-012:1272012:1272930 [7] NCCL INFO NVLS Head 7: 7 15 cnode7-012:1272012:1272930 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] 0/-1/-1->7->6 [2] 0/-1/-1->7->6 [3] 0/-1/-1->7->6 [4] 0/-1/-1->7->6 [5] 0/-1/-1->7->6 [6] 0/-1/-1->7->6 [7] 0/15/-1->7->-1 [8] -1/-1/-1->7->6 [9] 0/-1/-1->7->6 [10] 0/-1/-1->7->6 [11] 0/-1/-1->7->6 [12] 0/-1/-1->7->6 [13] 0/-1/-1->7->6 [14] 0/-1/-1->7->6 [15] 0/-1/-1->7->15 cnode7-012:1272012:1272930 [7] NCCL INFO P2P Chunksize set to 131072 cnode7-012:1272007:1272925 [2] NCCL INFO comm 0x55558061a960 rank 2 nRanks 16 nNodes 2 localRanks 8 localRank 2 MNNVL 0 cnode7-012:1272007:1272925 [2] NCCL INFO NVLS Head 0: 0 8 cnode7-012:1272007:1272925 [2] NCCL INFO NVLS Head 1: 1 9 cnode7-012:1272007:1272925 [2] NCCL INFO NVLS Head 2: 2 10 cnode7-012:1272007:1272925 [2] NCCL INFO NVLS Head 3: 3 11 cnode7-012:1272007:1272925 [2] NCCL INFO NVLS Head 4: 4 12 cnode7-012:1272007:1272925 [2] NCCL INFO NVLS Head 5: 5 13 cnode7-012:1272007:1272925 [2] NCCL INFO NVLS Head 6: 6 14 cnode7-012:1272007:1272925 [2] NCCL INFO NVLS Head 7: 7 15 cnode7-012:1272007:1272925 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/10/-1->2->-1 [3] -1/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->10 [11] -1/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 cnode7-012:1272007:1272925 [2] NCCL INFO P2P Chunksize set to 131072 cnode7-013:736483:737422 [2] NCCL INFO comm 0x55558161e2b0 rank 10 nRanks 16 nNodes 2 localRanks 8 localRank 2 MNNVL 0 cnode7-013:736483:737422 [2] NCCL INFO Trees [0] 11/-1/-1->10->9 [1] 11/-1/-1->10->9 [2] 11/-1/-1->10->2 [3] -1/-1/-1->10->9 [4] 11/-1/-1->10->9 [5] 11/-1/-1->10->9 [6] 11/-1/-1->10->9 [7] 11/-1/-1->10->9 [8] 11/-1/-1->10->9 [9] 11/-1/-1->10->9 [10] 11/2/-1->10->-1 [11] -1/-1/-1->10->9 [12] 11/-1/-1->10->9 [13] 11/-1/-1->10->9 [14] 11/-1/-1->10->9 [15] 11/-1/-1->10->9 cnode7-013:736483:737422 [2] NCCL INFO P2P Chunksize set to 131072 cnode7-012:1272011:1272928 [6] NCCL INFO comm 0x55557c81dd70 rank 6 nRanks 16 nNodes 2 localRanks 8 localRank 6 MNNVL 0 cnode7-012:1272011:1272928 [6] NCCL INFO NVLS Head 0: 0 8 cnode7-012:1272011:1272928 [6] NCCL INFO NVLS Head 1: 1 9 cnode7-012:1272011:1272928 [6] NCCL INFO NVLS Head 2: 2 10 cnode7-012:1272011:1272928 [6] NCCL INFO NVLS Head 3: 3 11 cnode7-012:1272011:1272928 [6] NCCL INFO NVLS Head 4: 4 12 cnode7-012:1272011:1272928 [6] NCCL INFO NVLS Head 5: 5 13 cnode7-012:1272011:1272928 [6] NCCL INFO NVLS Head 6: 6 14 cnode7-012:1272011:1272928 [6] NCCL INFO NVLS Head 7: 7 15 cnode7-012:1272011:1272928 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/14/-1->6->-1 [7] -1/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->14 [15] -1/-1/-1->6->5 cnode7-012:1272011:1272928 [6] NCCL INFO P2P Chunksize set to 131072 cnode7-012:1272007:1272925 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/CUMEM cnode7-012:1272007:1272925 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/CUMEM cnode7-012:1272007:1272925 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/CUMEM cnode7-012:1272007:1272925 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/CUMEM cnode7-012:1272007:1272925 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/CUMEM cnode7-012:1272007:1272925 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/CUMEM cnode7-012:1272009:1272927 [4] NCCL INFO comm 0x55558161a360 rank 4 nRanks 16 nNodes 2 localRanks 8 localRank 4 MNNVL 0 cnode7-012:1272009:1272927 [4] NCCL INFO NVLS Head 0: 0 8 cnode7-012:1272009:1272927 [4] NCCL INFO NVLS Head 1: 1 9 cnode7-012:1272009:1272927 [4] NCCL INFO NVLS Head 2: 2 10 cnode7-012:1272009:1272927 [4] NCCL INFO NVLS Head 3: 3 11 cnode7-012:1272009:1272927 [4] NCCL INFO NVLS Head 4: 4 12 cnode7-012:1272009:1272927 [4] NCCL INFO NVLS Head 5: 5 13 cnode7-012:1272009:1272927 [4] NCCL INFO NVLS Head 6: 6 14 cnode7-012:1272009:1272927 [4] NCCL INFO NVLS Head 7: 7 15 cnode7-012:1272009:1272927 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/12/-1->4->-1 [5] -1/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->12 [13] -1/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 cnode7-012:1272009:1272927 [4] NCCL INFO P2P Chunksize set to 131072 cnode7-013:736485:737373 [4] NCCL INFO Channel 00/0 : 12[4] -> 13[5] via P2P/CUMEM cnode7-013:736485:737373 [4] NCCL INFO Channel 02/0 : 12[4] -> 13[5] via P2P/CUMEM cnode7-013:736485:737373 [4] NCCL INFO Channel 06/0 : 12[4] -> 13[5] via P2P/CUMEM cnode7-013:736485:737373 [4] NCCL INFO Channel 08/0 : 12[4] -> 13[5] via P2P/CUMEM cnode7-013:736485:737373 [4] NCCL INFO Channel 10/0 : 12[4] -> 13[5] via P2P/CUMEM cnode7-013:736485:737373 [4] NCCL INFO Channel 14/0 : 12[4] -> 13[5] via P2P/CUMEM cnode7-013:736485:737373 [4] NCCL INFO Channel 05/0 : 4[4] -> 12[4] [receive] via NET/IB/4/GDRDMA cnode7-013:736485:737373 [4] NCCL INFO Channel 13/0 : 4[4] -> 12[4] [receive] via NET/IB/4/GDRDMA cnode7-013:736485:737373 [4] NCCL INFO Channel 04/0 : 12[4] -> 4[4] [send] via NET/IB/4/GDRDMA cnode7-013:736485:737373 [4] NCCL INFO Channel 12/0 : 12[4] -> 4[4] [send] via NET/IB/4/GDRDMA cnode7-013:736486:737423 [5] NCCL INFO Channel 00/0 : 13[5] -> 14[6] via P2P/CUMEM cnode7-013:736486:737423 [5] NCCL INFO Channel 02/0 : 13[5] -> 14[6] via P2P/CUMEM cnode7-013:736486:737423 [5] NCCL INFO Channel 04/0 : 13[5] -> 14[6] via P2P/CUMEM cnode7-013:736486:737423 [5] NCCL INFO Channel 06/0 : 13[5] -> 14[6] via P2P/CUMEM cnode7-013:736486:737423 [5] NCCL INFO Channel 08/0 : 13[5] -> 14[6] via P2P/CUMEM cnode7-013:736486:737423 [5] NCCL INFO Channel 10/0 : 13[5] -> 14[6] via P2P/CUMEM cnode7-013:736486:737423 [5] NCCL INFO Channel 12/0 : 13[5] -> 14[6] via P2P/CUMEM cnode7-013:736486:737423 [5] NCCL INFO Channel 14/0 : 13[5] -> 14[6] via P2P/CUMEM cnode7-013:736486:737423 [5] NCCL INFO Channel 04/0 : 5[5] -> 13[5] [receive] via NET/IB/6/GDRDMA cnode7-013:736486:737423 [5] NCCL INFO Channel 12/0 : 5[5] -> 13[5] [receive] via NET/IB/6/GDRDMA cnode7-013:736486:737423 [5] NCCL INFO Channel 05/0 : 13[5] -> 5[5] [send] via NET/IB/6/GDRDMA cnode7-013:736486:737423 [5] NCCL INFO Channel 13/0 : 13[5] -> 5[5] [send] via NET/IB/6/GDRDMA cnode7-013:736481:737376 [0] NCCL INFO Channel 02/0 : 8[0] -> 9[1] via P2P/CUMEM cnode7-013:736481:737376 [0] NCCL INFO Channel 04/0 : 8[0] -> 9[1] via P2P/CUMEM cnode7-013:736481:737376 [0] NCCL INFO Channel 06/0 : 8[0] -> 9[1] via P2P/CUMEM cnode7-013:736481:737376 [0] NCCL INFO Channel 10/0 : 8[0] -> 9[1] via P2P/CUMEM cnode7-013:736481:737376 [0] NCCL INFO Channel 12/0 : 8[0] -> 9[1] via P2P/CUMEM cnode7-013:736481:737376 [0] NCCL INFO Channel 14/0 : 8[0] -> 9[1] via P2P/CUMEM cnode7-013:736481:737376 [0] NCCL INFO Channel 01/0 : 8[0] -> 15[7] via P2P/CUMEM cnode7-013:736481:737376 [0] NCCL INFO Channel 03/0 : 8[0] -> 15[7] via P2P/CUMEM cnode7-013:736481:737376 [0] NCCL INFO Channel 05/0 : 8[0] -> 15[7] via P2P/CUMEM cnode7-013:736481:737376 [0] NCCL INFO Channel 07/0 : 8[0] -> 15[7] via P2P/CUMEM cnode7-013:736481:737376 [0] NCCL INFO Channel 09/0 : 8[0] -> 15[7] via P2P/CUMEM cnode7-013:736481:737376 [0] NCCL INFO Channel 11/0 : 8[0] -> 15[7] via P2P/CUMEM cnode7-013:736481:737376 [0] NCCL INFO Channel 13/0 : 8[0] -> 15[7] via P2P/CUMEM cnode7-013:736481:737376 [0] NCCL INFO Channel 15/0 : 8[0] -> 15[7] via P2P/CUMEM cnode7-013:736481:737376 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [receive] via NET/IB/0/GDRDMA cnode7-013:736481:737376 [0] NCCL INFO Channel 09/0 : 0[0] -> 8[0] [receive] via NET/IB/0/GDRDMA cnode7-013:736481:737376 [0] NCCL INFO Channel 00/0 : 8[0] -> 0[0] [send] via NET/IB/0/GDRDMA cnode7-013:736481:737376 [0] NCCL INFO Channel 08/0 : 8[0] -> 0[0] [send] via NET/IB/0/GDRDMA cnode7-012:1272011:1272928 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/CUMEM cnode7-012:1272011:1272928 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/CUMEM cnode7-012:1272011:1272928 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/CUMEM cnode7-012:1272011:1272928 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/CUMEM cnode7-012:1272011:1272928 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/CUMEM cnode7-012:1272011:1272928 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/CUMEM cnode7-013:736483:737422 [2] NCCL INFO Channel 00/0 : 10[2] -> 11[3] via P2P/CUMEM cnode7-013:736483:737422 [2] NCCL INFO Channel 04/0 : 10[2] -> 11[3] via P2P/CUMEM cnode7-013:736483:737422 [2] NCCL INFO Channel 06/0 : 10[2] -> 11[3] via P2P/CUMEM cnode7-013:736483:737422 [2] NCCL INFO Channel 08/0 : 10[2] -> 11[3] via P2P/CUMEM cnode7-013:736483:737422 [2] NCCL INFO Channel 12/0 : 10[2] -> 11[3] via P2P/CUMEM cnode7-013:736483:737422 [2] NCCL INFO Channel 14/0 : 10[2] -> 11[3] via P2P/CUMEM cnode7-013:736483:737422 [2] NCCL INFO Channel 03/0 : 2[2] -> 10[2] [receive] via NET/IB/2/GDRDMA cnode7-013:736483:737422 [2] NCCL INFO Channel 11/0 : 2[2] -> 10[2] [receive] via NET/IB/2/GDRDMA cnode7-013:736483:737422 [2] NCCL INFO Channel 02/0 : 10[2] -> 2[2] [send] via NET/IB/2/GDRDMA cnode7-013:736483:737422 [2] NCCL INFO Channel 10/0 : 10[2] -> 2[2] [send] via NET/IB/2/GDRDMA cnode7-013:736483:737422 [2] NCCL INFO Channel 01/0 : 10[2] -> 9[1] via P2P/CUMEM cnode7-013:736483:737422 [2] NCCL INFO Channel 03/0 : 10[2] -> 9[1] via P2P/CUMEM cnode7-012:1272006:1272926 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/CUMEM cnode7-012:1272006:1272926 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/CUMEM cnode7-012:1272006:1272926 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/CUMEM cnode7-012:1272006:1272926 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/CUMEM cnode7-012:1272006:1272926 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/CUMEM cnode7-012:1272006:1272926 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/CUMEM cnode7-012:1272006:1272926 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/CUMEM cnode7-012:1272006:1272926 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/CUMEM cnode7-012:1272006:1272926 [1] NCCL INFO Channel 01/0 : 9[1] -> 1[1] [receive] via NET/IB/1/GDRDMA cnode7-012:1272006:1272926 [1] NCCL INFO Channel 09/0 : 9[1] -> 1[1] [receive] via NET/IB/1/GDRDMA cnode7-012:1272006:1272926 [1] NCCL INFO Channel 00/0 : 1[1] -> 9[1] [send] via NET/IB/1/GDRDMA cnode7-012:1272006:1272926 [1] NCCL INFO Channel 08/0 : 1[1] -> 9[1] [send] via NET/IB/1/GDRDMA cnode7-012:1272006:1272926 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/CUMEM cnode7-012:1272006:1272926 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/CUMEM cnode7-012:1272006:1272926 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/CUMEM cnode7-012:1272006:1272926 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/CUMEM cnode7-012:1272006:1272926 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/CUMEM cnode7-012:1272005:1272924 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/CUMEM cnode7-012:1272005:1272924 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/CUMEM cnode7-012:1272005:1272924 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/CUMEM cnode7-012:1272005:1272924 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/CUMEM cnode7-012:1272005:1272924 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/CUMEM cnode7-012:1272005:1272924 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/CUMEM cnode7-012:1272005:1272924 [0] NCCL INFO Channel 00/0 : 0[0] -> 7[7] via P2P/CUMEM cnode7-012:1272005:1272924 [0] NCCL INFO Channel 02/0 : 0[0] -> 7[7] via P2P/CUMEM cnode7-012:1272005:1272924 [0] NCCL INFO Channel 04/0 : 0[0] -> 7[7] via P2P/CUMEM cnode7-012:1272005:1272924 [0] NCCL INFO Channel 06/0 : 0[0] -> 7[7] via P2P/CUMEM cnode7-012:1272005:1272924 [0] NCCL INFO Channel 08/0 : 0[0] -> 7[7] via P2P/CUMEM cnode7-012:1272005:1272924 [0] NCCL INFO Channel 10/0 : 0[0] -> 7[7] via P2P/CUMEM cnode7-012:1272005:1272924 [0] NCCL INFO Channel 12/0 : 0[0] -> 7[7] via P2P/CUMEM cnode7-012:1272005:1272924 [0] NCCL INFO Channel 14/0 : 0[0] -> 7[7] via P2P/CUMEM cnode7-012:1272005:1272924 [0] NCCL INFO Channel 00/0 : 8[0] -> 0[0] [receive] via NET/IB/0/GDRDMA cnode7-012:1272005:1272924 [0] NCCL INFO Channel 08/0 : 8[0] -> 0[0] [receive] via NET/IB/0/GDRDMA cnode7-012:1272005:1272924 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [send] via NET/IB/0/GDRDMA cnode7-012:1272005:1272924 [0] NCCL INFO Channel 09/0 : 0[0] -> 8[0] [send] via NET/IB/0/GDRDMA cnode7-012:1272010:1272998 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/CUMEM cnode7-012:1272010:1272998 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/CUMEM cnode7-012:1272010:1272998 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/CUMEM cnode7-012:1272010:1272998 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/CUMEM cnode7-012:1272010:1272998 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/CUMEM cnode7-012:1272010:1272998 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/CUMEM cnode7-012:1272010:1272998 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/CUMEM cnode7-012:1272010:1272998 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/CUMEM cnode7-012:1272010:1272998 [5] NCCL INFO Channel 05/0 : 13[5] -> 5[5] [receive] via NET/IB/6/GDRDMA cnode7-012:1272010:1272998 [5] NCCL INFO Channel 13/0 : 13[5] -> 5[5] [receive] via NET/IB/6/GDRDMA cnode7-012:1272010:1272998 [5] NCCL INFO Channel 04/0 : 5[5] -> 13[5] [send] via NET/IB/6/GDRDMA cnode7-012:1272010:1272998 [5] NCCL INFO Channel 12/0 : 5[5] -> 13[5] [send] via NET/IB/6/GDRDMA cnode7-013:736484:737409 [3] NCCL INFO Channel 00/0 : 11[3] -> 12[4] via P2P/CUMEM cnode7-013:736484:737409 [3] NCCL INFO Channel 02/0 : 11[3] -> 12[4] via P2P/CUMEM cnode7-013:736484:737409 [3] NCCL INFO Channel 04/0 : 11[3] -> 12[4] via P2P/CUMEM cnode7-013:736484:737409 [3] NCCL INFO Channel 06/0 : 11[3] -> 12[4] via P2P/CUMEM cnode7-013:736484:737409 [3] NCCL INFO Channel 08/0 : 11[3] -> 12[4] via P2P/CUMEM cnode7-013:736484:737409 [3] NCCL INFO Channel 10/0 : 11[3] -> 12[4] via P2P/CUMEM cnode7-013:736484:737409 [3] NCCL INFO Channel 12/0 : 11[3] -> 12[4] via P2P/CUMEM cnode7-013:736484:737409 [3] NCCL INFO Channel 14/0 : 11[3] -> 12[4] via P2P/CUMEM cnode7-013:736484:737409 [3] NCCL INFO Channel 02/0 : 3[3] -> 11[3] [receive] via NET/IB/3/GDRDMA cnode7-013:736484:737409 [3] NCCL INFO Channel 10/0 : 3[3] -> 11[3] [receive] via NET/IB/3/GDRDMA cnode7-013:736484:737409 [3] NCCL INFO Channel 03/0 : 11[3] -> 3[3] [send] via NET/IB/3/GDRDMA cnode7-013:736484:737409 [3] NCCL INFO Channel 11/0 : 11[3] -> 3[3] [send] via NET/IB/3/GDRDMA cnode7-013:736488:737372 [7] NCCL INFO Channel 06/0 : 7[7] -> 15[7] [receive] via NET/IB/8/GDRDMA cnode7-013:736488:737372 [7] NCCL INFO Channel 14/0 : 7[7] -> 15[7] [receive] via NET/IB/8/GDRDMA cnode7-013:736488:737372 [7] NCCL INFO Channel 07/0 : 15[7] -> 7[7] [send] via NET/IB/8/GDRDMA cnode7-013:736488:737372 [7] NCCL INFO Channel 15/0 : 15[7] -> 7[7] [send] via NET/IB/8/GDRDMA cnode7-013:736488:737372 [7] NCCL INFO Channel 00/0 : 15[7] -> 8[0] via P2P/CUMEM cnode7-013:736488:737372 [7] NCCL INFO Channel 02/0 : 15[7] -> 8[0] via P2P/CUMEM cnode7-013:736488:737372 [7] NCCL INFO Channel 04/0 : 15[7] -> 8[0] via P2P/CUMEM cnode7-013:736488:737372 [7] NCCL INFO Channel 06/0 : 15[7] -> 8[0] via P2P/CUMEM cnode7-013:736488:737372 [7] NCCL INFO Channel 08/0 : 15[7] -> 8[0] via P2P/CUMEM cnode7-013:736488:737372 [7] NCCL INFO Channel 10/0 : 15[7] -> 8[0] via P2P/CUMEM cnode7-013:736488:737372 [7] NCCL INFO Channel 12/0 : 15[7] -> 8[0] via P2P/CUMEM cnode7-013:736488:737372 [7] NCCL INFO Channel 14/0 : 15[7] -> 8[0] via P2P/CUMEM cnode7-013:736488:737372 [7] NCCL INFO Channel 01/0 : 15[7] -> 14[6] via P2P/CUMEM cnode7-013:736488:737372 [7] NCCL INFO Channel 03/0 : 15[7] -> 14[6] via P2P/CUMEM cnode7-013:736488:737372 [7] NCCL INFO Channel 05/0 : 15[7] -> 14[6] via P2P/CUMEM cnode7-013:736488:737372 [7] NCCL INFO Channel 09/0 : 15[7] -> 14[6] via P2P/CUMEM cnode7-013:736488:737372 [7] NCCL INFO Channel 11/0 : 15[7] -> 14[6] via P2P/CUMEM cnode7-013:736488:737372 [7] NCCL INFO Channel 13/0 : 15[7] -> 14[6] via P2P/CUMEM cnode7-013:736487:737367 [6] NCCL INFO Channel 00/0 : 14[6] -> 15[7] via P2P/CUMEM cnode7-013:736487:737367 [6] NCCL INFO Channel 02/0 : 14[6] -> 15[7] via P2P/CUMEM cnode7-013:736487:737367 [6] NCCL INFO Channel 04/0 : 14[6] -> 15[7] via P2P/CUMEM cnode7-013:736487:737367 [6] NCCL INFO Channel 08/0 : 14[6] -> 15[7] via P2P/CUMEM cnode7-013:736487:737367 [6] NCCL INFO Channel 10/0 : 14[6] -> 15[7] via P2P/CUMEM cnode7-013:736487:737367 [6] NCCL INFO Channel 12/0 : 14[6] -> 15[7] via P2P/CUMEM cnode7-013:736487:737367 [6] NCCL INFO Channel 07/0 : 6[6] -> 14[6] [receive] via NET/IB/7/GDRDMA cnode7-013:736487:737367 [6] NCCL INFO Channel 15/0 : 6[6] -> 14[6] [receive] via NET/IB/7/GDRDMA cnode7-013:736487:737367 [6] NCCL INFO Channel 06/0 : 14[6] -> 6[6] [send] via NET/IB/7/GDRDMA cnode7-013:736487:737367 [6] NCCL INFO Channel 14/0 : 14[6] -> 6[6] [send] via NET/IB/7/GDRDMA cnode7-013:736487:737367 [6] NCCL INFO Channel 01/0 : 14[6] -> 13[5] via P2P/CUMEM cnode7-013:736487:737367 [6] NCCL INFO Channel 03/0 : 14[6] -> 13[5] via P2P/CUMEM cnode7-013:736487:737367 [6] NCCL INFO Channel 05/0 : 14[6] -> 13[5] via P2P/CUMEM cnode7-013:736487:737367 [6] NCCL INFO Channel 07/0 : 14[6] -> 13[5] via P2P/CUMEM cnode7-013:736487:737367 [6] NCCL INFO Channel 09/0 : 14[6] -> 13[5] via P2P/CUMEM cnode7-013:736487:737367 [6] NCCL INFO Channel 11/0 : 14[6] -> 13[5] via P2P/CUMEM cnode7-013:736487:737367 [6] NCCL INFO Channel 13/0 : 14[6] -> 13[5] via P2P/CUMEM cnode7-013:736487:737367 [6] NCCL INFO Channel 15/0 : 14[6] -> 13[5] via P2P/CUMEM cnode7-012:1272008:1272929 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/CUMEM cnode7-012:1272008:1272929 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/CUMEM cnode7-012:1272008:1272929 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/CUMEM cnode7-012:1272008:1272929 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/CUMEM cnode7-012:1272008:1272929 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/CUMEM cnode7-012:1272008:1272929 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/CUMEM cnode7-012:1272008:1272929 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/CUMEM cnode7-012:1272008:1272929 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/CUMEM cnode7-012:1272008:1272929 [3] NCCL INFO Channel 03/0 : 11[3] -> 3[3] [receive] via NET/IB/3/GDRDMA cnode7-012:1272008:1272929 [3] NCCL INFO Channel 11/0 : 11[3] -> 3[3] [receive] via NET/IB/3/GDRDMA cnode7-012:1272008:1272929 [3] NCCL INFO Channel 02/0 : 3[3] -> 11[3] [send] via NET/IB/3/GDRDMA cnode7-012:1272008:1272929 [3] NCCL INFO Channel 10/0 : 3[3] -> 11[3] [send] via NET/IB/3/GDRDMA cnode7-012:1272008:1272929 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/CUMEM cnode7-012:1272008:1272929 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/CUMEM cnode7-012:1272008:1272929 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/CUMEM cnode7-012:1272008:1272929 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/CUMEM cnode7-012:1272008:1272929 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/CUMEM cnode7-012:1272008:1272929 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/CUMEM cnode7-013:736482:737436 [1] NCCL INFO Channel 00/0 : 9[1] -> 10[2] via P2P/CUMEM cnode7-013:736482:737436 [1] NCCL INFO Channel 02/0 : 9[1] -> 10[2] via P2P/CUMEM cnode7-013:736482:737436 [1] NCCL INFO Channel 04/0 : 9[1] -> 10[2] via P2P/CUMEM cnode7-013:736482:737436 [1] NCCL INFO Channel 06/0 : 9[1] -> 10[2] via P2P/CUMEM cnode7-013:736482:737436 [1] NCCL INFO Channel 08/0 : 9[1] -> 10[2] via P2P/CUMEM cnode7-013:736482:737436 [1] NCCL INFO Channel 10/0 : 9[1] -> 10[2] via P2P/CUMEM cnode7-013:736482:737436 [1] NCCL INFO Channel 12/0 : 9[1] -> 10[2] via P2P/CUMEM cnode7-013:736482:737436 [1] NCCL INFO Channel 14/0 : 9[1] -> 10[2] via P2P/CUMEM cnode7-013:736482:737436 [1] NCCL INFO Channel 00/0 : 1[1] -> 9[1] [receive] via NET/IB/1/GDRDMA cnode7-013:736482:737436 [1] NCCL INFO Channel 08/0 : 1[1] -> 9[1] [receive] via NET/IB/1/GDRDMA cnode7-013:736482:737436 [1] NCCL INFO Channel 01/0 : 9[1] -> 1[1] [send] via NET/IB/1/GDRDMA cnode7-013:736482:737436 [1] NCCL INFO Channel 09/0 : 9[1] -> 1[1] [send] via NET/IB/1/GDRDMA cnode7-013:736482:737436 [1] NCCL INFO Channel 03/0 : 9[1] -> 8[0] via P2P/CUMEM cnode7-013:736482:737436 [1] NCCL INFO Channel 05/0 : 9[1] -> 8[0] via P2P/CUMEM cnode7-013:736482:737436 [1] NCCL INFO Channel 07/0 : 9[1] -> 8[0] via P2P/CUMEM cnode7-013:736482:737436 [1] NCCL INFO Channel 11/0 : 9[1] -> 8[0] via P2P/CUMEM cnode7-013:736482:737436 [1] NCCL INFO Channel 13/0 : 9[1] -> 8[0] via P2P/CUMEM cnode7-013:736482:737436 [1] NCCL INFO Channel 15/0 : 9[1] -> 8[0] via P2P/CUMEM cnode7-012:1272007:1272925 [2] NCCL INFO Channel 02/0 : 10[2] -> 2[2] [receive] via NET/IB/2/GDRDMA cnode7-012:1272007:1272925 [2] NCCL INFO Channel 10/0 : 10[2] -> 2[2] [receive] via NET/IB/2/GDRDMA cnode7-012:1272007:1272925 [2] NCCL INFO Channel 03/0 : 2[2] -> 10[2] [send] via NET/IB/2/GDRDMA cnode7-012:1272007:1272925 [2] NCCL INFO Channel 11/0 : 2[2] -> 10[2] [send] via NET/IB/2/GDRDMA cnode7-012:1272007:1272925 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/CUMEM cnode7-012:1272007:1272925 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/CUMEM cnode7-012:1272007:1272925 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/CUMEM cnode7-012:1272007:1272925 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/CUMEM cnode7-012:1272007:1272925 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/CUMEM cnode7-012:1272007:1272925 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/CUMEM cnode7-012:1272007:1272925 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/CUMEM cnode7-012:1272007:1272925 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/CUMEM cnode7-013:736485:737373 [4] NCCL INFO Channel 01/0 : 12[4] -> 11[3] via P2P/CUMEM cnode7-013:736485:737373 [4] NCCL INFO Channel 03/0 : 12[4] -> 11[3] via P2P/CUMEM cnode7-013:736485:737373 [4] NCCL INFO Channel 05/0 : 12[4] -> 11[3] via P2P/CUMEM cnode7-013:736485:737373 [4] NCCL INFO Channel 07/0 : 12[4] -> 11[3] via P2P/CUMEM cnode7-013:736485:737373 [4] NCCL INFO Channel 09/0 : 12[4] -> 11[3] via P2P/CUMEM cnode7-013:736485:737373 [4] NCCL INFO Channel 11/0 : 12[4] -> 11[3] via P2P/CUMEM cnode7-013:736485:737373 [4] NCCL INFO Channel 13/0 : 12[4] -> 11[3] via P2P/CUMEM cnode7-013:736485:737373 [4] NCCL INFO Channel 15/0 : 12[4] -> 11[3] via P2P/CUMEM cnode7-013:736486:737423 [5] NCCL INFO Channel 01/0 : 13[5] -> 12[4] via P2P/CUMEM cnode7-013:736486:737423 [5] NCCL INFO Channel 03/0 : 13[5] -> 12[4] via P2P/CUMEM cnode7-013:736486:737423 [5] NCCL INFO Channel 07/0 : 13[5] -> 12[4] via P2P/CUMEM cnode7-013:736486:737423 [5] NCCL INFO Channel 09/0 : 13[5] -> 12[4] via P2P/CUMEM cnode7-013:736486:737423 [5] NCCL INFO Channel 11/0 : 13[5] -> 12[4] via P2P/CUMEM cnode7-013:736486:737423 [5] NCCL INFO Channel 15/0 : 13[5] -> 12[4] via P2P/CUMEM cnode7-012:1272012:1272930 [7] NCCL INFO Channel 07/0 : 15[7] -> 7[7] [receive] via NET/IB/8/GDRDMA cnode7-012:1272012:1272930 [7] NCCL INFO Channel 15/0 : 15[7] -> 7[7] [receive] via NET/IB/8/GDRDMA cnode7-012:1272012:1272930 [7] NCCL INFO Channel 06/0 : 7[7] -> 15[7] [send] via NET/IB/8/GDRDMA cnode7-012:1272012:1272930 [7] NCCL INFO Channel 14/0 : 7[7] -> 15[7] [send] via NET/IB/8/GDRDMA cnode7-012:1272012:1272930 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/CUMEM cnode7-012:1272012:1272930 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/CUMEM cnode7-012:1272012:1272930 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/CUMEM cnode7-012:1272012:1272930 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/CUMEM cnode7-012:1272012:1272930 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/CUMEM cnode7-012:1272012:1272930 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/CUMEM cnode7-012:1272012:1272930 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/CUMEM cnode7-012:1272012:1272930 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/CUMEM cnode7-012:1272012:1272930 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/CUMEM cnode7-012:1272012:1272930 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/CUMEM cnode7-012:1272012:1272930 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/CUMEM cnode7-012:1272012:1272930 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/CUMEM cnode7-012:1272012:1272930 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/CUMEM cnode7-012:1272012:1272930 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/CUMEM cnode7-012:1272009:1272927 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/CUMEM cnode7-012:1272009:1272927 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/CUMEM cnode7-012:1272009:1272927 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/CUMEM cnode7-012:1272009:1272927 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/CUMEM cnode7-012:1272009:1272927 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/CUMEM cnode7-012:1272009:1272927 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/CUMEM cnode7-012:1272009:1272927 [4] NCCL INFO Channel 04/0 : 12[4] -> 4[4] [receive] via NET/IB/4/GDRDMA cnode7-012:1272009:1272927 [4] NCCL INFO Channel 12/0 : 12[4] -> 4[4] [receive] via NET/IB/4/GDRDMA cnode7-012:1272009:1272927 [4] NCCL INFO Channel 05/0 : 4[4] -> 12[4] [send] via NET/IB/4/GDRDMA cnode7-012:1272009:1272927 [4] NCCL INFO Channel 13/0 : 4[4] -> 12[4] [send] via NET/IB/4/GDRDMA cnode7-012:1272009:1272927 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/CUMEM cnode7-012:1272009:1272927 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/CUMEM cnode7-012:1272009:1272927 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/CUMEM cnode7-012:1272009:1272927 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/CUMEM cnode7-012:1272009:1272927 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/CUMEM cnode7-012:1272009:1272927 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/CUMEM cnode7-012:1272009:1272927 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/CUMEM cnode7-012:1272009:1272927 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/CUMEM cnode7-012:1272011:1272928 [6] NCCL INFO Channel 06/0 : 14[6] -> 6[6] [receive] via NET/IB/7/GDRDMA cnode7-012:1272011:1272928 [6] NCCL INFO Channel 14/0 : 14[6] -> 6[6] [receive] via NET/IB/7/GDRDMA cnode7-012:1272011:1272928 [6] NCCL INFO Channel 07/0 : 6[6] -> 14[6] [send] via NET/IB/7/GDRDMA cnode7-012:1272011:1272928 [6] NCCL INFO Channel 15/0 : 6[6] -> 14[6] [send] via NET/IB/7/GDRDMA cnode7-012:1272011:1272928 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/CUMEM cnode7-012:1272011:1272928 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/CUMEM cnode7-012:1272011:1272928 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/CUMEM cnode7-012:1272011:1272928 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/CUMEM cnode7-012:1272011:1272928 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/CUMEM cnode7-012:1272011:1272928 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/CUMEM cnode7-012:1272011:1272928 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/CUMEM cnode7-012:1272011:1272928 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/CUMEM cnode7-013:736483:737422 [2] NCCL INFO Channel 05/0 : 10[2] -> 9[1] via P2P/CUMEM cnode7-013:736483:737422 [2] NCCL INFO Channel 07/0 : 10[2] -> 9[1] via P2P/CUMEM cnode7-013:736483:737422 [2] NCCL INFO Channel 09/0 : 10[2] -> 9[1] via P2P/CUMEM cnode7-013:736483:737422 [2] NCCL INFO Channel 11/0 : 10[2] -> 9[1] via P2P/CUMEM cnode7-013:736483:737422 [2] NCCL INFO Channel 13/0 : 10[2] -> 9[1] via P2P/CUMEM cnode7-013:736483:737422 [2] NCCL INFO Channel 15/0 : 10[2] -> 9[1] via P2P/CUMEM cnode7-012:1272006:1272926 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/CUMEM cnode7-012:1272010:1272998 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/CUMEM cnode7-012:1272010:1272998 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/CUMEM cnode7-012:1272010:1272998 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/CUMEM cnode7-012:1272010:1272998 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/CUMEM cnode7-012:1272010:1272998 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/CUMEM cnode7-012:1272010:1272998 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/CUMEM cnode7-013:736484:737409 [3] NCCL INFO Channel 01/0 : 11[3] -> 10[2] via P2P/CUMEM cnode7-013:736484:737409 [3] NCCL INFO Channel 05/0 : 11[3] -> 10[2] via P2P/CUMEM cnode7-013:736484:737409 [3] NCCL INFO Channel 07/0 : 11[3] -> 10[2] via P2P/CUMEM cnode7-013:736484:737409 [3] NCCL INFO Channel 09/0 : 11[3] -> 10[2] via P2P/CUMEM cnode7-013:736484:737409 [3] NCCL INFO Channel 13/0 : 11[3] -> 10[2] via P2P/CUMEM cnode7-013:736484:737409 [3] NCCL INFO Channel 15/0 : 11[3] -> 10[2] via P2P/CUMEM cnode7-012:1272007:1272925 [2] NCCL INFO Connected all rings cnode7-013:736483:737422 [2] NCCL INFO Connected all rings cnode7-013:736483:737422 [2] NCCL INFO Channel 01/0 : 10[2] -> 11[3] via P2P/CUMEM cnode7-013:736483:737422 [2] NCCL INFO Channel 02/0 : 10[2] -> 11[3] via P2P/CUMEM cnode7-013:736483:737422 [2] NCCL INFO Channel 05/0 : 10[2] -> 11[3] via P2P/CUMEM cnode7-013:736483:737422 [2] NCCL INFO Channel 07/0 : 10[2] -> 11[3] via P2P/CUMEM cnode7-013:736483:737422 [2] NCCL INFO Channel 09/0 : 10[2] -> 11[3] via P2P/CUMEM cnode7-012:1272006:1272926 [1] NCCL INFO Connected all rings cnode7-012:1272005:1272924 [0] NCCL INFO Connected all rings cnode7-012:1272005:1272924 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/CUMEM cnode7-012:1272005:1272924 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/CUMEM cnode7-012:1272005:1272924 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/CUMEM cnode7-012:1272005:1272924 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/CUMEM cnode7-012:1272005:1272924 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/CUMEM cnode7-012:1272005:1272924 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/CUMEM cnode7-012:1272005:1272924 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/CUMEM cnode7-012:1272005:1272924 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/CUMEM cnode7-012:1272010:1272998 [5] NCCL INFO Connected all rings cnode7-012:1272010:1272998 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/CUMEM cnode7-012:1272010:1272998 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/CUMEM cnode7-012:1272010:1272998 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/CUMEM cnode7-012:1272010:1272998 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/CUMEM cnode7-012:1272010:1272998 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/CUMEM cnode7-012:1272010:1272998 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/CUMEM cnode7-012:1272010:1272998 [5] NCCL INFO Channel 05/0 : 5[5] -> 13[5] [send] via NET/IB/6/GDRDMA cnode7-012:1272010:1272998 [5] NCCL INFO Channel 13/0 : 5[5] -> 13[5] [send] via NET/IB/6/GDRDMA cnode7-013:736484:737409 [3] NCCL INFO Connected all rings cnode7-013:736484:737409 [3] NCCL INFO Channel 01/0 : 11[3] -> 12[4] via P2P/CUMEM cnode7-013:736484:737409 [3] NCCL INFO Channel 03/0 : 11[3] -> 12[4] via P2P/CUMEM cnode7-013:736484:737409 [3] NCCL INFO Channel 05/0 : 11[3] -> 12[4] via P2P/CUMEM cnode7-013:736484:737409 [3] NCCL INFO Channel 07/0 : 11[3] -> 12[4] via P2P/CUMEM cnode7-013:736484:737409 [3] NCCL INFO Channel 09/0 : 11[3] -> 12[4] via P2P/CUMEM cnode7-013:736484:737409 [3] NCCL INFO Channel 11/0 : 11[3] -> 12[4] via P2P/CUMEM cnode7-013:736484:737409 [3] NCCL INFO Channel 13/0 : 11[3] -> 12[4] via P2P/CUMEM cnode7-013:736484:737409 [3] NCCL INFO Channel 15/0 : 11[3] -> 12[4] via P2P/CUMEM cnode7-013:736484:737409 [3] NCCL INFO Channel 03/0 : 3[3] -> 11[3] [receive] via NET/IB/3/GDRDMA cnode7-013:736484:737409 [3] NCCL INFO Channel 11/0 : 3[3] -> 11[3] [receive] via NET/IB/3/GDRDMA cnode7-013:736484:737409 [3] NCCL INFO Channel 00/0 : 11[3] -> 10[2] via P2P/CUMEM cnode7-013:736484:737409 [3] NCCL INFO Channel 02/0 : 11[3] -> 10[2] via P2P/CUMEM cnode7-013:736484:737409 [3] NCCL INFO Channel 04/0 : 11[3] -> 10[2] via P2P/CUMEM cnode7-013:736484:737409 [3] NCCL INFO Channel 06/0 : 11[3] -> 10[2] via P2P/CUMEM cnode7-013:736488:737372 [7] NCCL INFO Connected all rings cnode7-013:736488:737372 [7] NCCL INFO Channel 07/0 : 7[7] -> 15[7] [receive] via NET/IB/8/GDRDMA cnode7-013:736488:737372 [7] NCCL INFO Channel 15/0 : 7[7] -> 15[7] [receive] via NET/IB/8/GDRDMA cnode7-013:736487:737367 [6] NCCL INFO Connected all rings cnode7-013:736487:737367 [6] NCCL INFO Channel 01/0 : 14[6] -> 15[7] via P2P/CUMEM cnode7-013:736487:737367 [6] NCCL INFO Channel 03/0 : 14[6] -> 15[7] via P2P/CUMEM cnode7-013:736487:737367 [6] NCCL INFO Channel 05/0 : 14[6] -> 15[7] via P2P/CUMEM cnode7-013:736487:737367 [6] NCCL INFO Channel 06/0 : 14[6] -> 15[7] via P2P/CUMEM cnode7-013:736487:737367 [6] NCCL INFO Channel 09/0 : 14[6] -> 15[7] via P2P/CUMEM cnode7-013:736487:737367 [6] NCCL INFO Channel 11/0 : 14[6] -> 15[7] via P2P/CUMEM cnode7-013:736487:737367 [6] NCCL INFO Channel 13/0 : 14[6] -> 15[7] via P2P/CUMEM cnode7-013:736487:737367 [6] NCCL INFO Channel 14/0 : 14[6] -> 15[7] via P2P/CUMEM cnode7-013:736487:737367 [6] NCCL INFO Channel 06/0 : 6[6] -> 14[6] [receive] via NET/IB/7/GDRDMA cnode7-013:736487:737367 [6] NCCL INFO Channel 14/0 : 6[6] -> 14[6] [receive] via NET/IB/7/GDRDMA cnode7-013:736487:737367 [6] NCCL INFO Channel 00/0 : 14[6] -> 13[5] via P2P/CUMEM cnode7-013:736487:737367 [6] NCCL INFO Channel 02/0 : 14[6] -> 13[5] via P2P/CUMEM cnode7-013:736487:737367 [6] NCCL INFO Channel 04/0 : 14[6] -> 13[5] via P2P/CUMEM cnode7-013:736487:737367 [6] NCCL INFO Channel 08/0 : 14[6] -> 13[5] via P2P/CUMEM cnode7-013:736487:737367 [6] NCCL INFO Channel 10/0 : 14[6] -> 13[5] via P2P/CUMEM cnode7-013:736487:737367 [6] NCCL INFO Channel 12/0 : 14[6] -> 13[5] via P2P/CUMEM cnode7-012:1272008:1272929 [3] NCCL INFO Connected all rings cnode7-012:1272008:1272929 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/CUMEM cnode7-012:1272008:1272929 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/CUMEM cnode7-012:1272008:1272929 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/CUMEM cnode7-012:1272008:1272929 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/CUMEM cnode7-012:1272008:1272929 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/CUMEM cnode7-012:1272008:1272929 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/CUMEM cnode7-012:1272008:1272929 [3] NCCL INFO Channel 03/0 : 3[3] -> 11[3] [send] via NET/IB/3/GDRDMA cnode7-012:1272008:1272929 [3] NCCL INFO Channel 11/0 : 3[3] -> 11[3] [send] via NET/IB/3/GDRDMA cnode7-012:1272008:1272929 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/CUMEM cnode7-012:1272008:1272929 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/CUMEM cnode7-012:1272008:1272929 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/CUMEM cnode7-012:1272008:1272929 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/CUMEM cnode7-012:1272008:1272929 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/CUMEM cnode7-012:1272008:1272929 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/CUMEM cnode7-012:1272008:1272929 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/CUMEM cnode7-012:1272008:1272929 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/CUMEM cnode7-013:736482:737436 [1] NCCL INFO Connected all rings cnode7-013:736482:737436 [1] NCCL INFO Channel 01/0 : 9[1] -> 10[2] via P2P/CUMEM cnode7-013:736482:737436 [1] NCCL INFO Channel 03/0 : 9[1] -> 10[2] via P2P/CUMEM cnode7-013:736482:737436 [1] NCCL INFO Channel 05/0 : 9[1] -> 10[2] via P2P/CUMEM cnode7-013:736482:737436 [1] NCCL INFO Channel 07/0 : 9[1] -> 10[2] via P2P/CUMEM cnode7-013:736482:737436 [1] NCCL INFO Channel 09/0 : 9[1] -> 10[2] via P2P/CUMEM cnode7-013:736482:737436 [1] NCCL INFO Channel 11/0 : 9[1] -> 10[2] via P2P/CUMEM cnode7-013:736482:737436 [1] NCCL INFO Channel 13/0 : 9[1] -> 10[2] via P2P/CUMEM cnode7-013:736482:737436 [1] NCCL INFO Channel 15/0 : 9[1] -> 10[2] via P2P/CUMEM cnode7-013:736482:737436 [1] NCCL INFO Channel 01/0 : 1[1] -> 9[1] [receive] via NET/IB/1/GDRDMA cnode7-013:736482:737436 [1] NCCL INFO Channel 09/0 : 1[1] -> 9[1] [receive] via NET/IB/1/GDRDMA cnode7-013:736482:737436 [1] NCCL INFO Channel 00/0 : 9[1] -> 8[0] via P2P/CUMEM cnode7-013:736482:737436 [1] NCCL INFO Channel 02/0 : 9[1] -> 8[0] via P2P/CUMEM cnode7-013:736482:737436 [1] NCCL INFO Channel 04/0 : 9[1] -> 8[0] via P2P/CUMEM cnode7-013:736482:737436 [1] NCCL INFO Channel 06/0 : 9[1] -> 8[0] via P2P/CUMEM cnode7-013:736482:737436 [1] NCCL INFO Channel 08/0 : 9[1] -> 8[0] via P2P/CUMEM cnode7-013:736482:737436 [1] NCCL INFO Channel 10/0 : 9[1] -> 8[0] via P2P/CUMEM cnode7-013:736482:737436 [1] NCCL INFO Channel 12/0 : 9[1] -> 8[0] via P2P/CUMEM cnode7-013:736482:737436 [1] NCCL INFO Channel 14/0 : 9[1] -> 8[0] via P2P/CUMEM cnode7-012:1272007:1272925 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/CUMEM cnode7-012:1272007:1272925 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/CUMEM cnode7-012:1272007:1272925 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/CUMEM cnode7-012:1272007:1272925 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/CUMEM cnode7-012:1272007:1272925 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/CUMEM cnode7-012:1272007:1272925 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/CUMEM cnode7-012:1272007:1272925 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/CUMEM cnode7-012:1272007:1272925 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/CUMEM cnode7-012:1272007:1272925 [2] NCCL INFO Channel 02/0 : 2[2] -> 10[2] [send] via NET/IB/2/GDRDMA cnode7-012:1272007:1272925 [2] NCCL INFO Channel 10/0 : 2[2] -> 10[2] [send] via NET/IB/2/GDRDMA cnode7-012:1272007:1272925 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/CUMEM cnode7-012:1272007:1272925 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/CUMEM cnode7-012:1272007:1272925 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/CUMEM cnode7-012:1272007:1272925 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/CUMEM cnode7-012:1272007:1272925 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/CUMEM cnode7-012:1272007:1272925 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/CUMEM cnode7-012:1272007:1272925 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/CUMEM cnode7-012:1272007:1272925 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/CUMEM cnode7-013:736485:737373 [4] NCCL INFO Connected all rings cnode7-013:736485:737373 [4] NCCL INFO Channel 01/0 : 12[4] -> 13[5] via P2P/CUMEM cnode7-013:736485:737373 [4] NCCL INFO Channel 03/0 : 12[4] -> 13[5] via P2P/CUMEM cnode7-013:736485:737373 [4] NCCL INFO Channel 04/0 : 12[4] -> 13[5] via P2P/CUMEM cnode7-013:736485:737373 [4] NCCL INFO Channel 07/0 : 12[4] -> 13[5] via P2P/CUMEM cnode7-013:736485:737373 [4] NCCL INFO Channel 09/0 : 12[4] -> 13[5] via P2P/CUMEM cnode7-013:736485:737373 [4] NCCL INFO Channel 11/0 : 12[4] -> 13[5] via P2P/CUMEM cnode7-013:736485:737373 [4] NCCL INFO Channel 12/0 : 12[4] -> 13[5] via P2P/CUMEM cnode7-013:736485:737373 [4] NCCL INFO Channel 15/0 : 12[4] -> 13[5] via P2P/CUMEM cnode7-013:736485:737373 [4] NCCL INFO Channel 04/0 : 4[4] -> 12[4] [receive] via NET/IB/4/GDRDMA cnode7-013:736485:737373 [4] NCCL INFO Channel 12/0 : 4[4] -> 12[4] [receive] via NET/IB/4/GDRDMA cnode7-013:736485:737373 [4] NCCL INFO Channel 00/0 : 12[4] -> 11[3] via P2P/CUMEM cnode7-013:736485:737373 [4] NCCL INFO Channel 02/0 : 12[4] -> 11[3] via P2P/CUMEM cnode7-013:736485:737373 [4] NCCL INFO Channel 06/0 : 12[4] -> 11[3] via P2P/CUMEM cnode7-013:736485:737373 [4] NCCL INFO Channel 08/0 : 12[4] -> 11[3] via P2P/CUMEM cnode7-013:736485:737373 [4] NCCL INFO Channel 10/0 : 12[4] -> 11[3] via P2P/CUMEM cnode7-013:736485:737373 [4] NCCL INFO Channel 14/0 : 12[4] -> 11[3] via P2P/CUMEM cnode7-013:736486:737423 [5] NCCL INFO Connected all rings cnode7-013:736486:737423 [5] NCCL INFO Channel 01/0 : 13[5] -> 14[6] via P2P/CUMEM cnode7-013:736486:737423 [5] NCCL INFO Channel 03/0 : 13[5] -> 14[6] via P2P/CUMEM cnode7-013:736486:737423 [5] NCCL INFO Channel 05/0 : 13[5] -> 14[6] via P2P/CUMEM cnode7-013:736486:737423 [5] NCCL INFO Channel 07/0 : 13[5] -> 14[6] via P2P/CUMEM cnode7-013:736486:737423 [5] NCCL INFO Channel 09/0 : 13[5] -> 14[6] via P2P/CUMEM cnode7-013:736486:737423 [5] NCCL INFO Channel 11/0 : 13[5] -> 14[6] via P2P/CUMEM cnode7-013:736486:737423 [5] NCCL INFO Channel 13/0 : 13[5] -> 14[6] via P2P/CUMEM cnode7-013:736486:737423 [5] NCCL INFO Channel 15/0 : 13[5] -> 14[6] via P2P/CUMEM cnode7-013:736486:737423 [5] NCCL INFO Channel 05/0 : 5[5] -> 13[5] [receive] via NET/IB/6/GDRDMA cnode7-013:736486:737423 [5] NCCL INFO Channel 13/0 : 5[5] -> 13[5] [receive] via NET/IB/6/GDRDMA cnode7-013:736486:737423 [5] NCCL INFO Channel 00/0 : 13[5] -> 12[4] via P2P/CUMEM cnode7-013:736486:737423 [5] NCCL INFO Channel 02/0 : 13[5] -> 12[4] via P2P/CUMEM cnode7-013:736486:737423 [5] NCCL INFO Channel 04/0 : 13[5] -> 12[4] via P2P/CUMEM cnode7-013:736486:737423 [5] NCCL INFO Channel 06/0 : 13[5] -> 12[4] via P2P/CUMEM cnode7-013:736486:737423 [5] NCCL INFO Channel 08/0 : 13[5] -> 12[4] via P2P/CUMEM cnode7-013:736486:737423 [5] NCCL INFO Channel 10/0 : 13[5] -> 12[4] via P2P/CUMEM cnode7-013:736486:737423 [5] NCCL INFO Channel 12/0 : 13[5] -> 12[4] via P2P/CUMEM cnode7-013:736486:737423 [5] NCCL INFO Channel 14/0 : 13[5] -> 12[4] via P2P/CUMEM cnode7-012:1272012:1272930 [7] NCCL INFO Connected all rings cnode7-012:1272012:1272930 [7] NCCL INFO Channel 07/0 : 7[7] -> 15[7] [send] via NET/IB/8/GDRDMA cnode7-012:1272012:1272930 [7] NCCL INFO Channel 15/0 : 7[7] -> 15[7] [send] via NET/IB/8/GDRDMA cnode7-012:1272012:1272930 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/CUMEM cnode7-012:1272012:1272930 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/CUMEM cnode7-012:1272012:1272930 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/CUMEM cnode7-012:1272012:1272930 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/CUMEM cnode7-012:1272012:1272930 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/CUMEM cnode7-012:1272012:1272930 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/CUMEM cnode7-012:1272012:1272930 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/CUMEM cnode7-012:1272012:1272930 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/CUMEM cnode7-012:1272012:1272930 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/CUMEM cnode7-012:1272012:1272930 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/CUMEM cnode7-012:1272012:1272930 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/CUMEM cnode7-012:1272012:1272930 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/CUMEM cnode7-012:1272012:1272930 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/CUMEM cnode7-012:1272012:1272930 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/CUMEM cnode7-013:736481:737376 [0] NCCL INFO Connected all rings cnode7-013:736481:737376 [0] NCCL INFO Channel 00/0 : 8[0] -> 9[1] via P2P/CUMEM cnode7-013:736481:737376 [0] NCCL INFO Channel 03/0 : 8[0] -> 9[1] via P2P/CUMEM cnode7-013:736481:737376 [0] NCCL INFO Channel 05/0 : 8[0] -> 9[1] via P2P/CUMEM cnode7-013:736481:737376 [0] NCCL INFO Channel 07/0 : 8[0] -> 9[1] via P2P/CUMEM cnode7-013:736481:737376 [0] NCCL INFO Channel 08/0 : 8[0] -> 9[1] via P2P/CUMEM cnode7-013:736481:737376 [0] NCCL INFO Channel 11/0 : 8[0] -> 9[1] via P2P/CUMEM cnode7-013:736481:737376 [0] NCCL INFO Channel 13/0 : 8[0] -> 9[1] via P2P/CUMEM cnode7-013:736481:737376 [0] NCCL INFO Channel 15/0 : 8[0] -> 9[1] via P2P/CUMEM cnode7-013:736481:737376 [0] NCCL INFO Channel 02/0 : 8[0] -> 15[7] via P2P/CUMEM cnode7-013:736481:737376 [0] NCCL INFO Channel 04/0 : 8[0] -> 15[7] via P2P/CUMEM cnode7-013:736481:737376 [0] NCCL INFO Channel 06/0 : 8[0] -> 15[7] via P2P/CUMEM cnode7-013:736481:737376 [0] NCCL INFO Channel 10/0 : 8[0] -> 15[7] via P2P/CUMEM cnode7-013:736481:737376 [0] NCCL INFO Channel 12/0 : 8[0] -> 15[7] via P2P/CUMEM cnode7-013:736481:737376 [0] NCCL INFO Channel 14/0 : 8[0] -> 15[7] via P2P/CUMEM cnode7-013:736481:737376 [0] NCCL INFO Channel 00/0 : 0[0] -> 8[0] [receive] via NET/IB/0/GDRDMA cnode7-013:736481:737376 [0] NCCL INFO Channel 08/0 : 0[0] -> 8[0] [receive] via NET/IB/0/GDRDMA cnode7-012:1272009:1272927 [4] NCCL INFO Connected all rings cnode7-012:1272009:1272927 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/CUMEM cnode7-012:1272009:1272927 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/CUMEM cnode7-012:1272009:1272927 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/CUMEM cnode7-012:1272009:1272927 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/CUMEM cnode7-012:1272009:1272927 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/CUMEM cnode7-012:1272009:1272927 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/CUMEM cnode7-012:1272009:1272927 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/CUMEM cnode7-012:1272009:1272927 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/CUMEM cnode7-012:1272009:1272927 [4] NCCL INFO Channel 04/0 : 4[4] -> 12[4] [send] via NET/IB/4/GDRDMA cnode7-012:1272009:1272927 [4] NCCL INFO Channel 12/0 : 4[4] -> 12[4] [send] via NET/IB/4/GDRDMA cnode7-012:1272009:1272927 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/CUMEM cnode7-012:1272009:1272927 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/CUMEM cnode7-012:1272009:1272927 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/CUMEM cnode7-012:1272009:1272927 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/CUMEM cnode7-012:1272009:1272927 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/CUMEM cnode7-012:1272009:1272927 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/CUMEM cnode7-012:1272009:1272927 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/CUMEM cnode7-012:1272009:1272927 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/CUMEM cnode7-012:1272011:1272928 [6] NCCL INFO Connected all rings cnode7-012:1272011:1272928 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/CUMEM cnode7-012:1272011:1272928 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/CUMEM cnode7-012:1272011:1272928 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/CUMEM cnode7-012:1272011:1272928 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/CUMEM cnode7-012:1272011:1272928 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/CUMEM cnode7-012:1272011:1272928 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/CUMEM cnode7-012:1272011:1272928 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/CUMEM cnode7-012:1272011:1272928 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/CUMEM cnode7-012:1272011:1272928 [6] NCCL INFO Channel 06/0 : 6[6] -> 14[6] [send] via NET/IB/7/GDRDMA cnode7-012:1272011:1272928 [6] NCCL INFO Channel 14/0 : 6[6] -> 14[6] [send] via NET/IB/7/GDRDMA cnode7-012:1272011:1272928 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/CUMEM cnode7-012:1272011:1272928 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/CUMEM cnode7-012:1272011:1272928 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/CUMEM cnode7-012:1272011:1272928 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/CUMEM cnode7-012:1272011:1272928 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/CUMEM cnode7-012:1272011:1272928 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/CUMEM cnode7-012:1272011:1272928 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/CUMEM cnode7-012:1272011:1272928 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/CUMEM cnode7-013:736483:737422 [2] NCCL INFO Channel 10/0 : 10[2] -> 11[3] via P2P/CUMEM cnode7-013:736483:737422 [2] NCCL INFO Channel 13/0 : 10[2] -> 11[3] via P2P/CUMEM cnode7-013:736483:737422 [2] NCCL INFO Channel 15/0 : 10[2] -> 11[3] via P2P/CUMEM cnode7-013:736483:737422 [2] NCCL INFO Channel 02/0 : 2[2] -> 10[2] [receive] via NET/IB/2/GDRDMA cnode7-013:736483:737422 [2] NCCL INFO Channel 10/0 : 2[2] -> 10[2] [receive] via NET/IB/2/GDRDMA cnode7-013:736483:737422 [2] NCCL INFO Channel 00/0 : 10[2] -> 9[1] via P2P/CUMEM cnode7-013:736483:737422 [2] NCCL INFO Channel 04/0 : 10[2] -> 9[1] via P2P/CUMEM cnode7-013:736483:737422 [2] NCCL INFO Channel 06/0 : 10[2] -> 9[1] via P2P/CUMEM cnode7-013:736483:737422 [2] NCCL INFO Channel 08/0 : 10[2] -> 9[1] via P2P/CUMEM cnode7-013:736483:737422 [2] NCCL INFO Channel 12/0 : 10[2] -> 9[1] via P2P/CUMEM cnode7-013:736483:737422 [2] NCCL INFO Channel 14/0 : 10[2] -> 9[1] via P2P/CUMEM cnode7-012:1272006:1272926 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/CUMEM cnode7-012:1272006:1272926 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/CUMEM cnode7-012:1272006:1272926 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/CUMEM cnode7-012:1272006:1272926 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/CUMEM cnode7-012:1272006:1272926 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/CUMEM cnode7-012:1272006:1272926 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/CUMEM cnode7-012:1272006:1272926 [1] NCCL INFO Channel 01/0 : 1[1] -> 9[1] [send] via NET/IB/1/GDRDMA cnode7-012:1272006:1272926 [1] NCCL INFO Channel 09/0 : 1[1] -> 9[1] [send] via NET/IB/1/GDRDMA cnode7-012:1272006:1272926 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/CUMEM cnode7-012:1272006:1272926 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/CUMEM cnode7-012:1272006:1272926 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/CUMEM cnode7-012:1272006:1272926 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/CUMEM cnode7-012:1272006:1272926 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/CUMEM cnode7-012:1272006:1272926 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/CUMEM cnode7-012:1272006:1272926 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/CUMEM cnode7-012:1272006:1272926 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/CUMEM cnode7-012:1272005:1272924 [0] NCCL INFO Channel 01/0 : 0[0] -> 7[7] via P2P/CUMEM cnode7-012:1272005:1272924 [0] NCCL INFO Channel 03/0 : 0[0] -> 7[7] via P2P/CUMEM cnode7-012:1272005:1272924 [0] NCCL INFO Channel 05/0 : 0[0] -> 7[7] via P2P/CUMEM cnode7-012:1272005:1272924 [0] NCCL INFO Channel 07/0 : 0[0] -> 7[7] via P2P/CUMEM cnode7-012:1272005:1272924 [0] NCCL INFO Channel 09/0 : 0[0] -> 7[7] via P2P/CUMEM cnode7-012:1272005:1272924 [0] NCCL INFO Channel 11/0 : 0[0] -> 7[7] via P2P/CUMEM cnode7-012:1272005:1272924 [0] NCCL INFO Channel 13/0 : 0[0] -> 7[7] via P2P/CUMEM cnode7-012:1272005:1272924 [0] NCCL INFO Channel 15/0 : 0[0] -> 7[7] via P2P/CUMEM cnode7-012:1272005:1272924 [0] NCCL INFO Channel 00/0 : 0[0] -> 8[0] [send] via NET/IB/0/GDRDMA cnode7-012:1272005:1272924 [0] NCCL INFO Channel 08/0 : 0[0] -> 8[0] [send] via NET/IB/0/GDRDMA cnode7-012:1272010:1272998 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/CUMEM cnode7-012:1272010:1272998 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/CUMEM cnode7-012:1272010:1272998 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/CUMEM cnode7-012:1272010:1272998 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/CUMEM cnode7-012:1272010:1272998 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/CUMEM cnode7-012:1272010:1272998 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/CUMEM cnode7-012:1272010:1272998 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/CUMEM cnode7-012:1272010:1272998 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/CUMEM cnode7-013:736484:737409 [3] NCCL INFO Channel 08/0 : 11[3] -> 10[2] via P2P/CUMEM cnode7-013:736484:737409 [3] NCCL INFO Channel 10/0 : 11[3] -> 10[2] via P2P/CUMEM cnode7-013:736484:737409 [3] NCCL INFO Channel 12/0 : 11[3] -> 10[2] via P2P/CUMEM cnode7-013:736484:737409 [3] NCCL INFO Channel 14/0 : 11[3] -> 10[2] via P2P/CUMEM cnode7-013:736488:737372 [7] NCCL INFO Channel 01/0 : 15[7] -> 8[0] via P2P/CUMEM cnode7-013:736488:737372 [7] NCCL INFO Channel 03/0 : 15[7] -> 8[0] via P2P/CUMEM cnode7-013:736488:737372 [7] NCCL INFO Channel 05/0 : 15[7] -> 8[0] via P2P/CUMEM cnode7-013:736488:737372 [7] NCCL INFO Channel 07/0 : 15[7] -> 8[0] via P2P/CUMEM cnode7-013:736488:737372 [7] NCCL INFO Channel 09/0 : 15[7] -> 8[0] via P2P/CUMEM cnode7-013:736488:737372 [7] NCCL INFO Channel 11/0 : 15[7] -> 8[0] via P2P/CUMEM cnode7-013:736488:737372 [7] NCCL INFO Channel 13/0 : 15[7] -> 8[0] via P2P/CUMEM cnode7-013:736488:737372 [7] NCCL INFO Channel 15/0 : 15[7] -> 8[0] via P2P/CUMEM cnode7-013:736488:737372 [7] NCCL INFO Channel 00/0 : 15[7] -> 14[6] via P2P/CUMEM cnode7-013:736488:737372 [7] NCCL INFO Channel 02/0 : 15[7] -> 14[6] via P2P/CUMEM cnode7-013:736488:737372 [7] NCCL INFO Channel 04/0 : 15[7] -> 14[6] via P2P/CUMEM cnode7-013:736488:737372 [7] NCCL INFO Channel 06/0 : 15[7] -> 14[6] via P2P/CUMEM cnode7-013:736488:737372 [7] NCCL INFO Channel 08/0 : 15[7] -> 14[6] via P2P/CUMEM cnode7-013:736488:737372 [7] NCCL INFO Channel 10/0 : 15[7] -> 14[6] via P2P/CUMEM cnode7-013:736488:737372 [7] NCCL INFO Channel 12/0 : 15[7] -> 14[6] via P2P/CUMEM cnode7-013:736488:737372 [7] NCCL INFO Channel 14/0 : 15[7] -> 14[6] via P2P/CUMEM cnode7-013:736483:737422 [2] NCCL INFO Connected all trees cnode7-012:1272007:1272925 [2] NCCL INFO Connected all trees cnode7-013:736483:737422 [2] NCCL INFO NVLS comm 0x55558161e2b0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736 cnode7-013:736488:737372 [7] NCCL INFO Connected all trees cnode7-013:736481:737376 [0] NCCL INFO Connected all trees cnode7-012:1272011:1272928 [6] NCCL INFO Connected all trees cnode7-012:1272008:1272929 [3] NCCL INFO Connected all trees cnode7-012:1272009:1272927 [4] NCCL INFO Connected all trees cnode7-012:1272010:1272998 [5] NCCL INFO Connected all trees cnode7-013:736482:737436 [1] NCCL INFO Connected all trees cnode7-013:736484:737409 [3] NCCL INFO Connected all trees cnode7-013:736487:737367 [6] NCCL INFO Connected all trees cnode7-013:736486:737423 [5] NCCL INFO Connected all trees cnode7-012:1272005:1272924 [0] NCCL INFO Connected all trees cnode7-012:1272012:1272930 [7] NCCL INFO Connected all trees cnode7-012:1272006:1272926 [1] NCCL INFO Connected all trees cnode7-013:736488:737372 [7] NCCL INFO NVLS comm 0x555560c15d00 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736 cnode7-013:736485:737373 [4] NCCL INFO Connected all trees cnode7-013:736487:737367 [6] NCCL INFO NVLS comm 0x555561a1a9e0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736 cnode7-012:1272008:1272929 [3] NCCL INFO NVLS comm 0x55557b81cee0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736 cnode7-013:736482:737436 [1] NCCL INFO NVLS comm 0x555580e21bb0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736 cnode7-012:1272007:1272925 [2] NCCL INFO NVLS comm 0x55558061a960 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736 cnode7-013:736485:737373 [4] NCCL INFO NVLS comm 0x555560c19b70 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736 cnode7-013:736486:737423 [5] NCCL INFO NVLS comm 0x555581618ce0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736 cnode7-012:1272012:1272930 [7] NCCL INFO NVLS comm 0x55557b820a90 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736 cnode7-013:736481:737376 [0] NCCL INFO NVLS comm 0x555563b6a610 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736 cnode7-012:1272009:1272927 [4] NCCL INFO NVLS comm 0x55558161a360 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736 cnode7-012:1272011:1272928 [6] NCCL INFO NVLS comm 0x55557c81dd70 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736 cnode7-012:1272006:1272926 [1] NCCL INFO NVLS comm 0x55558161ded0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736 cnode7-012:1272005:1272924 [0] NCCL INFO NVLS comm 0x555563bd8350 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736 cnode7-012:1272010:1272998 [5] NCCL INFO NVLS comm 0x555572620330 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736 cnode7-013:736484:737409 [3] NCCL INFO NVLS comm 0x55557b419c80 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736 cnode7-013:736484:737409 [3] NCCL INFO Channel 00/0 : 3[3] -> 11[3] [receive] via NET/IB/3/GDRDMA cnode7-013:736484:737409 [3] NCCL INFO Channel 01/0 : 3[3] -> 11[3] [receive] via NET/IB/3/GDRDMA cnode7-013:736484:737409 [3] NCCL INFO Channel 04/0 : 3[3] -> 11[3] [receive] via NET/IB/3/GDRDMA cnode7-013:736484:737409 [3] NCCL INFO Channel 05/0 : 3[3] -> 11[3] [receive] via NET/IB/3/GDRDMA cnode7-013:736484:737409 [3] NCCL INFO Channel 06/0 : 3[3] -> 11[3] [receive] via NET/IB/3/GDRDMA cnode7-013:736484:737409 [3] NCCL INFO Channel 07/0 : 3[3] -> 11[3] [receive] via NET/IB/3/GDRDMA cnode7-013:736484:737409 [3] NCCL INFO Channel 08/0 : 3[3] -> 11[3] [receive] via NET/IB/3/GDRDMA cnode7-013:736484:737409 [3] NCCL INFO Channel 09/0 : 3[3] -> 11[3] [receive] via NET/IB/3/GDRDMA cnode7-013:736488:737372 [7] NCCL INFO Channel 00/0 : 7[7] -> 15[7] [receive] via NET/IB/8/GDRDMA cnode7-013:736488:737372 [7] NCCL INFO Channel 01/0 : 7[7] -> 15[7] [receive] via NET/IB/8/GDRDMA cnode7-013:736488:737372 [7] NCCL INFO Channel 02/0 : 7[7] -> 15[7] [receive] via NET/IB/8/GDRDMA cnode7-013:736488:737372 [7] NCCL INFO Channel 03/0 : 7[7] -> 15[7] [receive] via NET/IB/8/GDRDMA cnode7-013:736488:737372 [7] NCCL INFO Channel 04/0 : 7[7] -> 15[7] [receive] via NET/IB/8/GDRDMA cnode7-013:736488:737372 [7] NCCL INFO Channel 05/0 : 7[7] -> 15[7] [receive] via NET/IB/8/GDRDMA cnode7-013:736488:737372 [7] NCCL INFO Channel 08/0 : 7[7] -> 15[7] [receive] via NET/IB/8/GDRDMA cnode7-013:736488:737372 [7] NCCL INFO Channel 09/0 : 7[7] -> 15[7] [receive] via NET/IB/8/GDRDMA cnode7-013:736487:737367 [6] NCCL INFO Channel 00/0 : 6[6] -> 14[6] [receive] via NET/IB/7/GDRDMA cnode7-013:736487:737367 [6] NCCL INFO Channel 01/0 : 6[6] -> 14[6] [receive] via NET/IB/7/GDRDMA cnode7-013:736487:737367 [6] NCCL INFO Channel 02/0 : 6[6] -> 14[6] [receive] via NET/IB/7/GDRDMA cnode7-013:736487:737367 [6] NCCL INFO Channel 03/0 : 6[6] -> 14[6] [receive] via NET/IB/7/GDRDMA cnode7-013:736487:737367 [6] NCCL INFO Channel 04/0 : 6[6] -> 14[6] [receive] via NET/IB/7/GDRDMA cnode7-013:736487:737367 [6] NCCL INFO Channel 05/0 : 6[6] -> 14[6] [receive] via NET/IB/7/GDRDMA cnode7-013:736487:737367 [6] NCCL INFO Channel 08/0 : 6[6] -> 14[6] [receive] via NET/IB/7/GDRDMA cnode7-013:736487:737367 [6] NCCL INFO Channel 09/0 : 6[6] -> 14[6] [receive] via NET/IB/7/GDRDMA cnode7-012:1272008:1272929 [3] NCCL INFO Channel 00/0 : 11[3] -> 3[3] [receive] via NET/IB/3/GDRDMA cnode7-012:1272008:1272929 [3] NCCL INFO Channel 01/0 : 11[3] -> 3[3] [receive] via NET/IB/3/GDRDMA cnode7-012:1272008:1272929 [3] NCCL INFO Channel 02/0 : 11[3] -> 3[3] [receive] via NET/IB/3/GDRDMA cnode7-012:1272008:1272929 [3] NCCL INFO Channel 04/0 : 11[3] -> 3[3] [receive] via NET/IB/3/GDRDMA cnode7-012:1272008:1272929 [3] NCCL INFO Channel 05/0 : 11[3] -> 3[3] [receive] via NET/IB/3/GDRDMA cnode7-012:1272008:1272929 [3] NCCL INFO Channel 06/0 : 11[3] -> 3[3] [receive] via NET/IB/3/GDRDMA cnode7-012:1272008:1272929 [3] NCCL INFO Channel 07/0 : 11[3] -> 3[3] [receive] via NET/IB/3/GDRDMA cnode7-012:1272008:1272929 [3] NCCL INFO Channel 08/0 : 11[3] -> 3[3] [receive] via NET/IB/3/GDRDMA cnode7-012:1272007:1272925 [2] NCCL INFO Channel 00/0 : 10[2] -> 2[2] [receive] via NET/IB/2/GDRDMA cnode7-012:1272007:1272925 [2] NCCL INFO Channel 01/0 : 10[2] -> 2[2] [receive] via NET/IB/2/GDRDMA cnode7-012:1272007:1272925 [2] NCCL INFO Channel 03/0 : 10[2] -> 2[2] [receive] via NET/IB/2/GDRDMA cnode7-012:1272007:1272925 [2] NCCL INFO Channel 04/0 : 10[2] -> 2[2] [receive] via NET/IB/2/GDRDMA cnode7-012:1272007:1272925 [2] NCCL INFO Channel 05/0 : 10[2] -> 2[2] [receive] via NET/IB/2/GDRDMA cnode7-012:1272007:1272925 [2] NCCL INFO Channel 06/0 : 10[2] -> 2[2] [receive] via NET/IB/2/GDRDMA cnode7-012:1272007:1272925 [2] NCCL INFO Channel 07/0 : 10[2] -> 2[2] [receive] via NET/IB/2/GDRDMA cnode7-012:1272007:1272925 [2] NCCL INFO Channel 08/0 : 10[2] -> 2[2] [receive] via NET/IB/2/GDRDMA cnode7-013:736482:737436 [1] NCCL INFO Channel 02/0 : 1[1] -> 9[1] [receive] via NET/IB/1/GDRDMA cnode7-013:736482:737436 [1] NCCL INFO Channel 03/0 : 1[1] -> 9[1] [receive] via NET/IB/1/GDRDMA cnode7-013:736482:737436 [1] NCCL INFO Channel 04/0 : 1[1] -> 9[1] [receive] via NET/IB/1/GDRDMA cnode7-013:736482:737436 [1] NCCL INFO Channel 05/0 : 1[1] -> 9[1] [receive] via NET/IB/1/GDRDMA cnode7-013:736482:737436 [1] NCCL INFO Channel 06/0 : 1[1] -> 9[1] [receive] via NET/IB/1/GDRDMA cnode7-013:736482:737436 [1] NCCL INFO Channel 07/0 : 1[1] -> 9[1] [receive] via NET/IB/1/GDRDMA cnode7-013:736482:737436 [1] NCCL INFO Channel 10/0 : 1[1] -> 9[1] [receive] via NET/IB/1/GDRDMA cnode7-013:736482:737436 [1] NCCL INFO Channel 11/0 : 1[1] -> 9[1] [receive] via NET/IB/1/GDRDMA cnode7-013:736485:737373 [4] NCCL INFO Channel 00/0 : 4[4] -> 12[4] [receive] via NET/IB/4/GDRDMA cnode7-013:736485:737373 [4] NCCL INFO Channel 01/0 : 4[4] -> 12[4] [receive] via NET/IB/4/GDRDMA cnode7-013:736485:737373 [4] NCCL INFO Channel 02/0 : 4[4] -> 12[4] [receive] via NET/IB/4/GDRDMA cnode7-013:736485:737373 [4] NCCL INFO Channel 03/0 : 4[4] -> 12[4] [receive] via NET/IB/4/GDRDMA cnode7-013:736485:737373 [4] NCCL INFO Channel 06/0 : 4[4] -> 12[4] [receive] via NET/IB/4/GDRDMA cnode7-013:736485:737373 [4] NCCL INFO Channel 07/0 : 4[4] -> 12[4] [receive] via NET/IB/4/GDRDMA cnode7-013:736485:737373 [4] NCCL INFO Channel 08/0 : 4[4] -> 12[4] [receive] via NET/IB/4/GDRDMA cnode7-013:736485:737373 [4] NCCL INFO Channel 09/0 : 4[4] -> 12[4] [receive] via NET/IB/4/GDRDMA cnode7-013:736486:737423 [5] NCCL INFO Channel 00/0 : 5[5] -> 13[5] [receive] via NET/IB/6/GDRDMA cnode7-013:736486:737423 [5] NCCL INFO Channel 01/0 : 5[5] -> 13[5] [receive] via NET/IB/6/GDRDMA cnode7-013:736486:737423 [5] NCCL INFO Channel 02/0 : 5[5] -> 13[5] [receive] via NET/IB/6/GDRDMA cnode7-013:736486:737423 [5] NCCL INFO Channel 03/0 : 5[5] -> 13[5] [receive] via NET/IB/6/GDRDMA cnode7-013:736486:737423 [5] NCCL INFO Channel 06/0 : 5[5] -> 13[5] [receive] via NET/IB/6/GDRDMA cnode7-013:736486:737423 [5] NCCL INFO Channel 07/0 : 5[5] -> 13[5] [receive] via NET/IB/6/GDRDMA cnode7-013:736486:737423 [5] NCCL INFO Channel 08/0 : 5[5] -> 13[5] [receive] via NET/IB/6/GDRDMA cnode7-013:736486:737423 [5] NCCL INFO Channel 09/0 : 5[5] -> 13[5] [receive] via NET/IB/6/GDRDMA cnode7-013:736481:737376 [0] NCCL INFO Channel 02/0 : 0[0] -> 8[0] [receive] via NET/IB/0/GDRDMA cnode7-013:736481:737376 [0] NCCL INFO Channel 03/0 : 0[0] -> 8[0] [receive] via NET/IB/0/GDRDMA cnode7-013:736481:737376 [0] NCCL INFO Channel 04/0 : 0[0] -> 8[0] [receive] via NET/IB/0/GDRDMA cnode7-013:736481:737376 [0] NCCL INFO Channel 05/0 : 0[0] -> 8[0] [receive] via NET/IB/0/GDRDMA cnode7-013:736481:737376 [0] NCCL INFO Channel 06/0 : 0[0] -> 8[0] [receive] via NET/IB/0/GDRDMA cnode7-013:736481:737376 [0] NCCL INFO Channel 07/0 : 0[0] -> 8[0] [receive] via NET/IB/0/GDRDMA cnode7-013:736481:737376 [0] NCCL INFO Channel 10/0 : 0[0] -> 8[0] [receive] via NET/IB/0/GDRDMA cnode7-013:736481:737376 [0] NCCL INFO Channel 11/0 : 0[0] -> 8[0] [receive] via NET/IB/0/GDRDMA cnode7-013:736481:737376 [0] NCCL INFO Channel 12/0 : 0[0] -> 8[0] [receive] via NET/IB/0/GDRDMA cnode7-013:736483:737422 [2] NCCL INFO Channel 00/0 : 2[2] -> 10[2] [receive] via NET/IB/2/GDRDMA cnode7-013:736483:737422 [2] NCCL INFO Channel 01/0 : 2[2] -> 10[2] [receive] via NET/IB/2/GDRDMA cnode7-013:736483:737422 [2] NCCL INFO Channel 04/0 : 2[2] -> 10[2] [receive] via NET/IB/2/GDRDMA cnode7-013:736483:737422 [2] NCCL INFO Channel 05/0 : 2[2] -> 10[2] [receive] via NET/IB/2/GDRDMA cnode7-013:736483:737422 [2] NCCL INFO Channel 06/0 : 2[2] -> 10[2] [receive] via NET/IB/2/GDRDMA cnode7-013:736483:737422 [2] NCCL INFO Channel 07/0 : 2[2] -> 10[2] [receive] via NET/IB/2/GDRDMA cnode7-013:736483:737422 [2] NCCL INFO Channel 08/0 : 2[2] -> 10[2] [receive] via NET/IB/2/GDRDMA cnode7-013:736483:737422 [2] NCCL INFO Channel 09/0 : 2[2] -> 10[2] [receive] via NET/IB/2/GDRDMA cnode7-013:736483:737422 [2] NCCL INFO Channel 12/0 : 2[2] -> 10[2] [receive] via NET/IB/2/GDRDMA cnode7-013:736483:737422 [2] NCCL INFO Channel 13/0 : 2[2] -> 10[2] [receive] via NET/IB/2/GDRDMA cnode7-013:736483:737422 [2] NCCL INFO Channel 14/0 : 2[2] -> 10[2] [receive] via NET/IB/2/GDRDMA cnode7-013:736483:737422 [2] NCCL INFO Channel 15/0 : 2[2] -> 10[2] [receive] via NET/IB/2/GDRDMA cnode7-013:736483:737422 [2] NCCL INFO Channel 00/0 : 10[2] -> 2[2] [send] via NET/IB/2/GDRDMA cnode7-013:736483:737422 [2] NCCL INFO Channel 01/0 : 10[2] -> 2[2] [send] via NET/IB/2/GDRDMA cnode7-013:736483:737422 [2] NCCL INFO Channel 03/0 : 10[2] -> 2[2] [send] via NET/IB/2/GDRDMA cnode7-013:736483:737422 [2] NCCL INFO Channel 04/0 : 10[2] -> 2[2] [send] via NET/IB/2/GDRDMA cnode7-013:736483:737422 [2] NCCL INFO Channel 05/0 : 10[2] -> 2[2] [send] via NET/IB/2/GDRDMA cnode7-013:736483:737422 [2] NCCL INFO Channel 06/0 : 10[2] -> 2[2] [send] via NET/IB/2/GDRDMA cnode7-013:736483:737422 [2] NCCL INFO Channel 07/0 : 10[2] -> 2[2] [send] via NET/IB/2/GDRDMA cnode7-013:736483:737422 [2] NCCL INFO Channel 08/0 : 10[2] -> 2[2] [send] via NET/IB/2/GDRDMA cnode7-013:736483:737422 [2] NCCL INFO Channel 09/0 : 10[2] -> 2[2] [send] via NET/IB/2/GDRDMA cnode7-013:736483:737422 [2] NCCL INFO Channel 11/0 : 10[2] -> 2[2] [send] via NET/IB/2/GDRDMA cnode7-013:736483:737422 [2] NCCL INFO Channel 12/0 : 10[2] -> 2[2] [send] via NET/IB/2/GDRDMA cnode7-013:736483:737422 [2] NCCL INFO Channel 13/0 : 10[2] -> 2[2] [send] via NET/IB/2/GDRDMA cnode7-013:736483:737422 [2] NCCL INFO Channel 14/0 : 10[2] -> 2[2] [send] via NET/IB/2/GDRDMA cnode7-013:736483:737422 [2] NCCL INFO Channel 15/0 : 10[2] -> 2[2] [send] via NET/IB/2/GDRDMA cnode7-012:1272012:1272930 [7] NCCL INFO Channel 00/0 : 15[7] -> 7[7] [receive] via NET/IB/8/GDRDMA cnode7-012:1272012:1272930 [7] NCCL INFO Channel 01/0 : 15[7] -> 7[7] [receive] via NET/IB/8/GDRDMA cnode7-012:1272012:1272930 [7] NCCL INFO Channel 02/0 : 15[7] -> 7[7] [receive] via NET/IB/8/GDRDMA cnode7-012:1272012:1272930 [7] NCCL INFO Channel 03/0 : 15[7] -> 7[7] [receive] via NET/IB/8/GDRDMA cnode7-012:1272012:1272930 [7] NCCL INFO Channel 04/0 : 15[7] -> 7[7] [receive] via NET/IB/8/GDRDMA cnode7-012:1272012:1272930 [7] NCCL INFO Channel 05/0 : 15[7] -> 7[7] [receive] via NET/IB/8/GDRDMA cnode7-012:1272012:1272930 [7] NCCL INFO Channel 06/0 : 15[7] -> 7[7] [receive] via NET/IB/8/GDRDMA cnode7-012:1272012:1272930 [7] NCCL INFO Channel 08/0 : 15[7] -> 7[7] [receive] via NET/IB/8/GDRDMA cnode7-012:1272012:1272930 [7] NCCL INFO Channel 09/0 : 15[7] -> 7[7] [receive] via NET/IB/8/GDRDMA cnode7-012:1272012:1272930 [7] NCCL INFO Channel 10/0 : 15[7] -> 7[7] [receive] via NET/IB/8/GDRDMA cnode7-012:1272012:1272930 [7] NCCL INFO Channel 11/0 : 15[7] -> 7[7] [receive] via NET/IB/8/GDRDMA cnode7-012:1272012:1272930 [7] NCCL INFO Channel 12/0 : 15[7] -> 7[7] [receive] via NET/IB/8/GDRDMA cnode7-012:1272012:1272930 [7] NCCL INFO Channel 13/0 : 15[7] -> 7[7] [receive] via NET/IB/8/GDRDMA cnode7-012:1272012:1272930 [7] NCCL INFO Channel 14/0 : 15[7] -> 7[7] [receive] via NET/IB/8/GDRDMA cnode7-012:1272012:1272930 [7] NCCL INFO Channel 00/0 : 7[7] -> 15[7] [send] via NET/IB/8/GDRDMA cnode7-012:1272012:1272930 [7] NCCL INFO Channel 01/0 : 7[7] -> 15[7] [send] via NET/IB/8/GDRDMA cnode7-012:1272012:1272930 [7] NCCL INFO Channel 02/0 : 7[7] -> 15[7] [send] via NET/IB/8/GDRDMA cnode7-012:1272012:1272930 [7] NCCL INFO Channel 03/0 : 7[7] -> 15[7] [send] via NET/IB/8/GDRDMA cnode7-012:1272009:1272927 [4] NCCL INFO Channel 00/0 : 12[4] -> 4[4] [receive] via NET/IB/4/GDRDMA cnode7-012:1272009:1272927 [4] NCCL INFO Channel 01/0 : 12[4] -> 4[4] [receive] via NET/IB/4/GDRDMA cnode7-012:1272009:1272927 [4] NCCL INFO Channel 02/0 : 12[4] -> 4[4] [receive] via NET/IB/4/GDRDMA cnode7-012:1272009:1272927 [4] NCCL INFO Channel 03/0 : 12[4] -> 4[4] [receive] via NET/IB/4/GDRDMA cnode7-012:1272009:1272927 [4] NCCL INFO Channel 05/0 : 12[4] -> 4[4] [receive] via NET/IB/4/GDRDMA cnode7-012:1272009:1272927 [4] NCCL INFO Channel 06/0 : 12[4] -> 4[4] [receive] via NET/IB/4/GDRDMA cnode7-012:1272009:1272927 [4] NCCL INFO Channel 07/0 : 12[4] -> 4[4] [receive] via NET/IB/4/GDRDMA cnode7-012:1272009:1272927 [4] NCCL INFO Channel 08/0 : 12[4] -> 4[4] [receive] via NET/IB/4/GDRDMA cnode7-012:1272009:1272927 [4] NCCL INFO Channel 09/0 : 12[4] -> 4[4] [receive] via NET/IB/4/GDRDMA cnode7-012:1272009:1272927 [4] NCCL INFO Channel 10/0 : 12[4] -> 4[4] [receive] via NET/IB/4/GDRDMA cnode7-012:1272009:1272927 [4] NCCL INFO Channel 11/0 : 12[4] -> 4[4] [receive] via NET/IB/4/GDRDMA cnode7-012:1272009:1272927 [4] NCCL INFO Channel 13/0 : 12[4] -> 4[4] [receive] via NET/IB/4/GDRDMA cnode7-012:1272009:1272927 [4] NCCL INFO Channel 14/0 : 12[4] -> 4[4] [receive] via NET/IB/4/GDRDMA cnode7-012:1272009:1272927 [4] NCCL INFO Channel 15/0 : 12[4] -> 4[4] [receive] via NET/IB/4/GDRDMA cnode7-012:1272009:1272927 [4] NCCL INFO Channel 00/0 : 4[4] -> 12[4] [send] via NET/IB/4/GDRDMA cnode7-012:1272009:1272927 [4] NCCL INFO Channel 01/0 : 4[4] -> 12[4] [send] via NET/IB/4/GDRDMA cnode7-012:1272009:1272927 [4] NCCL INFO Channel 02/0 : 4[4] -> 12[4] [send] via NET/IB/4/GDRDMA cnode7-012:1272009:1272927 [4] NCCL INFO Channel 03/0 : 4[4] -> 12[4] [send] via NET/IB/4/GDRDMA cnode7-012:1272009:1272927 [4] NCCL INFO Channel 06/0 : 4[4] -> 12[4] [send] via NET/IB/4/GDRDMA cnode7-012:1272009:1272927 [4] NCCL INFO Channel 07/0 : 4[4] -> 12[4] [send] via NET/IB/4/GDRDMA cnode7-012:1272009:1272927 [4] NCCL INFO Channel 08/0 : 4[4] -> 12[4] [send] via NET/IB/4/GDRDMA cnode7-012:1272009:1272927 [4] NCCL INFO Channel 09/0 : 4[4] -> 12[4] [send] via NET/IB/4/GDRDMA cnode7-012:1272009:1272927 [4] NCCL INFO Channel 10/0 : 4[4] -> 12[4] [send] via NET/IB/4/GDRDMA cnode7-012:1272009:1272927 [4] NCCL INFO Channel 11/0 : 4[4] -> 12[4] [send] via NET/IB/4/GDRDMA cnode7-012:1272009:1272927 [4] NCCL INFO Channel 14/0 : 4[4] -> 12[4] [send] via NET/IB/4/GDRDMA cnode7-012:1272009:1272927 [4] NCCL INFO Channel 15/0 : 4[4] -> 12[4] [send] via NET/IB/4/GDRDMA cnode7-012:1272011:1272928 [6] NCCL INFO Channel 00/0 : 14[6] -> 6[6] [receive] via NET/IB/7/GDRDMA cnode7-012:1272011:1272928 [6] NCCL INFO Channel 01/0 : 14[6] -> 6[6] [receive] via NET/IB/7/GDRDMA cnode7-012:1272011:1272928 [6] NCCL INFO Channel 02/0 : 14[6] -> 6[6] [receive] via NET/IB/7/GDRDMA cnode7-012:1272011:1272928 [6] NCCL INFO Channel 03/0 : 14[6] -> 6[6] [receive] via NET/IB/7/GDRDMA cnode7-012:1272011:1272928 [6] NCCL INFO Channel 04/0 : 14[6] -> 6[6] [receive] via NET/IB/7/GDRDMA cnode7-012:1272011:1272928 [6] NCCL INFO Channel 05/0 : 14[6] -> 6[6] [receive] via NET/IB/7/GDRDMA cnode7-012:1272011:1272928 [6] NCCL INFO Channel 07/0 : 14[6] -> 6[6] [receive] via NET/IB/7/GDRDMA cnode7-012:1272011:1272928 [6] NCCL INFO Channel 08/0 : 14[6] -> 6[6] [receive] via NET/IB/7/GDRDMA cnode7-012:1272011:1272928 [6] NCCL INFO Channel 09/0 : 14[6] -> 6[6] [receive] via NET/IB/7/GDRDMA cnode7-012:1272011:1272928 [6] NCCL INFO Channel 10/0 : 14[6] -> 6[6] [receive] via NET/IB/7/GDRDMA cnode7-012:1272011:1272928 [6] NCCL INFO Channel 11/0 : 14[6] -> 6[6] [receive] via NET/IB/7/GDRDMA cnode7-012:1272011:1272928 [6] NCCL INFO Channel 12/0 : 14[6] -> 6[6] [receive] via NET/IB/7/GDRDMA cnode7-012:1272011:1272928 [6] NCCL INFO Channel 13/0 : 14[6] -> 6[6] [receive] via NET/IB/7/GDRDMA cnode7-012:1272011:1272928 [6] NCCL INFO Channel 15/0 : 14[6] -> 6[6] [receive] via NET/IB/7/GDRDMA cnode7-012:1272011:1272928 [6] NCCL INFO Channel 00/0 : 6[6] -> 14[6] [send] via NET/IB/7/GDRDMA cnode7-012:1272011:1272928 [6] NCCL INFO Channel 01/0 : 6[6] -> 14[6] [send] via NET/IB/7/GDRDMA cnode7-012:1272011:1272928 [6] NCCL INFO Channel 02/0 : 6[6] -> 14[6] [send] via NET/IB/7/GDRDMA cnode7-012:1272011:1272928 [6] NCCL INFO Channel 03/0 : 6[6] -> 14[6] [send] via NET/IB/7/GDRDMA cnode7-012:1272011:1272928 [6] NCCL INFO Channel 04/0 : 6[6] -> 14[6] [send] via NET/IB/7/GDRDMA cnode7-012:1272011:1272928 [6] NCCL INFO Channel 05/0 : 6[6] -> 14[6] [send] via NET/IB/7/GDRDMA cnode7-012:1272011:1272928 [6] NCCL INFO Channel 08/0 : 6[6] -> 14[6] [send] via NET/IB/7/GDRDMA cnode7-012:1272011:1272928 [6] NCCL INFO Channel 09/0 : 6[6] -> 14[6] [send] via NET/IB/7/GDRDMA cnode7-012:1272011:1272928 [6] NCCL INFO Channel 10/0 : 6[6] -> 14[6] [send] via NET/IB/7/GDRDMA cnode7-012:1272011:1272928 [6] NCCL INFO Channel 11/0 : 6[6] -> 14[6] [send] via NET/IB/7/GDRDMA cnode7-012:1272011:1272928 [6] NCCL INFO Channel 12/0 : 6[6] -> 14[6] [send] via NET/IB/7/GDRDMA cnode7-012:1272011:1272928 [6] NCCL INFO Channel 13/0 : 6[6] -> 14[6] [send] via NET/IB/7/GDRDMA cnode7-012:1272006:1272926 [1] NCCL INFO Channel 00/0 : 9[1] -> 1[1] [receive] via NET/IB/1/GDRDMA cnode7-012:1272006:1272926 [1] NCCL INFO Channel 02/0 : 9[1] -> 1[1] [receive] via NET/IB/1/GDRDMA cnode7-012:1272006:1272926 [1] NCCL INFO Channel 03/0 : 9[1] -> 1[1] [receive] via NET/IB/1/GDRDMA cnode7-012:1272006:1272926 [1] NCCL INFO Channel 04/0 : 9[1] -> 1[1] [receive] via NET/IB/1/GDRDMA cnode7-012:1272006:1272926 [1] NCCL INFO Channel 05/0 : 9[1] -> 1[1] [receive] via NET/IB/1/GDRDMA cnode7-012:1272006:1272926 [1] NCCL INFO Channel 06/0 : 9[1] -> 1[1] [receive] via NET/IB/1/GDRDMA cnode7-012:1272006:1272926 [1] NCCL INFO Channel 07/0 : 9[1] -> 1[1] [receive] via NET/IB/1/GDRDMA cnode7-012:1272006:1272926 [1] NCCL INFO Channel 08/0 : 9[1] -> 1[1] [receive] via NET/IB/1/GDRDMA cnode7-012:1272006:1272926 [1] NCCL INFO Channel 10/0 : 9[1] -> 1[1] [receive] via NET/IB/1/GDRDMA cnode7-012:1272006:1272926 [1] NCCL INFO Channel 11/0 : 9[1] -> 1[1] [receive] via NET/IB/1/GDRDMA cnode7-012:1272006:1272926 [1] NCCL INFO Channel 12/0 : 9[1] -> 1[1] [receive] via NET/IB/1/GDRDMA cnode7-012:1272006:1272926 [1] NCCL INFO Channel 13/0 : 9[1] -> 1[1] [receive] via NET/IB/1/GDRDMA cnode7-012:1272006:1272926 [1] NCCL INFO Channel 14/0 : 9[1] -> 1[1] [receive] via NET/IB/1/GDRDMA cnode7-012:1272006:1272926 [1] NCCL INFO Channel 15/0 : 9[1] -> 1[1] [receive] via NET/IB/1/GDRDMA cnode7-012:1272006:1272926 [1] NCCL INFO Channel 02/0 : 1[1] -> 9[1] [send] via NET/IB/1/GDRDMA cnode7-012:1272006:1272926 [1] NCCL INFO Channel 03/0 : 1[1] -> 9[1] [send] via NET/IB/1/GDRDMA cnode7-012:1272006:1272926 [1] NCCL INFO Channel 04/0 : 1[1] -> 9[1] [send] via NET/IB/1/GDRDMA cnode7-012:1272006:1272926 [1] NCCL INFO Channel 05/0 : 1[1] -> 9[1] [send] via NET/IB/1/GDRDMA cnode7-012:1272006:1272926 [1] NCCL INFO Channel 06/0 : 1[1] -> 9[1] [send] via NET/IB/1/GDRDMA cnode7-012:1272006:1272926 [1] NCCL INFO Channel 07/0 : 1[1] -> 9[1] [send] via NET/IB/1/GDRDMA cnode7-012:1272006:1272926 [1] NCCL INFO Channel 10/0 : 1[1] -> 9[1] [send] via NET/IB/1/GDRDMA cnode7-012:1272006:1272926 [1] NCCL INFO Channel 11/0 : 1[1] -> 9[1] [send] via NET/IB/1/GDRDMA cnode7-012:1272006:1272926 [1] NCCL INFO Channel 12/0 : 1[1] -> 9[1] [send] via NET/IB/1/GDRDMA cnode7-012:1272006:1272926 [1] NCCL INFO Channel 13/0 : 1[1] -> 9[1] [send] via NET/IB/1/GDRDMA cnode7-012:1272006:1272926 [1] NCCL INFO Channel 14/0 : 1[1] -> 9[1] [send] via NET/IB/1/GDRDMA cnode7-012:1272006:1272926 [1] NCCL INFO Channel 15/0 : 1[1] -> 9[1] [send] via NET/IB/1/GDRDMA cnode7-012:1272005:1272924 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [receive] via NET/IB/0/GDRDMA cnode7-012:1272005:1272924 [0] NCCL INFO Channel 02/0 : 8[0] -> 0[0] [receive] via NET/IB/0/GDRDMA cnode7-012:1272005:1272924 [0] NCCL INFO Channel 03/0 : 8[0] -> 0[0] [receive] via NET/IB/0/GDRDMA cnode7-012:1272005:1272924 [0] NCCL INFO Channel 04/0 : 8[0] -> 0[0] [receive] via NET/IB/0/GDRDMA cnode7-012:1272005:1272924 [0] NCCL INFO Channel 05/0 : 8[0] -> 0[0] [receive] via NET/IB/0/GDRDMA cnode7-012:1272005:1272924 [0] NCCL INFO Channel 06/0 : 8[0] -> 0[0] [receive] via NET/IB/0/GDRDMA cnode7-012:1272005:1272924 [0] NCCL INFO Channel 07/0 : 8[0] -> 0[0] [receive] via NET/IB/0/GDRDMA cnode7-012:1272005:1272924 [0] NCCL INFO Channel 09/0 : 8[0] -> 0[0] [receive] via NET/IB/0/GDRDMA cnode7-012:1272005:1272924 [0] NCCL INFO Channel 10/0 : 8[0] -> 0[0] [receive] via NET/IB/0/GDRDMA cnode7-012:1272005:1272924 [0] NCCL INFO Channel 11/0 : 8[0] -> 0[0] [receive] via NET/IB/0/GDRDMA cnode7-012:1272005:1272924 [0] NCCL INFO Channel 12/0 : 8[0] -> 0[0] [receive] via NET/IB/0/GDRDMA cnode7-012:1272005:1272924 [0] NCCL INFO Channel 13/0 : 8[0] -> 0[0] [receive] via NET/IB/0/GDRDMA cnode7-012:1272005:1272924 [0] NCCL INFO Channel 14/0 : 8[0] -> 0[0] [receive] via NET/IB/0/GDRDMA cnode7-012:1272005:1272924 [0] NCCL INFO Channel 15/0 : 8[0] -> 0[0] [receive] via NET/IB/0/GDRDMA cnode7-012:1272005:1272924 [0] NCCL INFO Channel 02/0 : 0[0] -> 8[0] [send] via NET/IB/0/GDRDMA cnode7-012:1272005:1272924 [0] NCCL INFO Channel 03/0 : 0[0] -> 8[0] [send] via NET/IB/0/GDRDMA cnode7-012:1272005:1272924 [0] NCCL INFO Channel 04/0 : 0[0] -> 8[0] [send] via NET/IB/0/GDRDMA cnode7-012:1272005:1272924 [0] NCCL INFO Channel 05/0 : 0[0] -> 8[0] [send] via NET/IB/0/GDRDMA cnode7-012:1272005:1272924 [0] NCCL INFO Channel 06/0 : 0[0] -> 8[0] [send] via NET/IB/0/GDRDMA cnode7-012:1272005:1272924 [0] NCCL INFO Channel 07/0 : 0[0] -> 8[0] [send] via NET/IB/0/GDRDMA cnode7-012:1272005:1272924 [0] NCCL INFO Channel 10/0 : 0[0] -> 8[0] [send] via NET/IB/0/GDRDMA cnode7-012:1272005:1272924 [0] NCCL INFO Channel 11/0 : 0[0] -> 8[0] [send] via NET/IB/0/GDRDMA cnode7-012:1272005:1272924 [0] NCCL INFO Channel 12/0 : 0[0] -> 8[0] [send] via NET/IB/0/GDRDMA cnode7-012:1272005:1272924 [0] NCCL INFO Channel 13/0 : 0[0] -> 8[0] [send] via NET/IB/0/GDRDMA cnode7-012:1272005:1272924 [0] NCCL INFO Channel 14/0 : 0[0] -> 8[0] [send] via NET/IB/0/GDRDMA cnode7-012:1272005:1272924 [0] NCCL INFO Channel 15/0 : 0[0] -> 8[0] [send] via NET/IB/0/GDRDMA cnode7-012:1272010:1272998 [5] NCCL INFO Channel 00/0 : 13[5] -> 5[5] [receive] via NET/IB/6/GDRDMA cnode7-012:1272010:1272998 [5] NCCL INFO Channel 01/0 : 13[5] -> 5[5] [receive] via NET/IB/6/GDRDMA cnode7-012:1272010:1272998 [5] NCCL INFO Channel 02/0 : 13[5] -> 5[5] [receive] via NET/IB/6/GDRDMA cnode7-012:1272010:1272998 [5] NCCL INFO Channel 03/0 : 13[5] -> 5[5] [receive] via NET/IB/6/GDRDMA cnode7-012:1272010:1272998 [5] NCCL INFO Channel 04/0 : 13[5] -> 5[5] [receive] via NET/IB/6/GDRDMA cnode7-012:1272010:1272998 [5] NCCL INFO Channel 06/0 : 13[5] -> 5[5] [receive] via NET/IB/6/GDRDMA cnode7-012:1272010:1272998 [5] NCCL INFO Channel 07/0 : 13[5] -> 5[5] [receive] via NET/IB/6/GDRDMA cnode7-012:1272010:1272998 [5] NCCL INFO Channel 08/0 : 13[5] -> 5[5] [receive] via NET/IB/6/GDRDMA cnode7-012:1272010:1272998 [5] NCCL INFO Channel 09/0 : 13[5] -> 5[5] [receive] via NET/IB/6/GDRDMA cnode7-012:1272010:1272998 [5] NCCL INFO Channel 10/0 : 13[5] -> 5[5] [receive] via NET/IB/6/GDRDMA cnode7-012:1272010:1272998 [5] NCCL INFO Channel 11/0 : 13[5] -> 5[5] [receive] via NET/IB/6/GDRDMA cnode7-012:1272010:1272998 [5] NCCL INFO Channel 12/0 : 13[5] -> 5[5] [receive] via NET/IB/6/GDRDMA cnode7-012:1272010:1272998 [5] NCCL INFO Channel 14/0 : 13[5] -> 5[5] [receive] via NET/IB/6/GDRDMA cnode7-012:1272010:1272998 [5] NCCL INFO Channel 15/0 : 13[5] -> 5[5] [receive] via NET/IB/6/GDRDMA cnode7-012:1272010:1272998 [5] NCCL INFO Channel 00/0 : 5[5] -> 13[5] [send] via NET/IB/6/GDRDMA cnode7-012:1272010:1272998 [5] NCCL INFO Channel 01/0 : 5[5] -> 13[5] [send] via NET/IB/6/GDRDMA cnode7-012:1272010:1272998 [5] NCCL INFO Channel 02/0 : 5[5] -> 13[5] [send] via NET/IB/6/GDRDMA cnode7-012:1272010:1272998 [5] NCCL INFO Channel 03/0 : 5[5] -> 13[5] [send] via NET/IB/6/GDRDMA cnode7-012:1272010:1272998 [5] NCCL INFO Channel 06/0 : 5[5] -> 13[5] [send] via NET/IB/6/GDRDMA cnode7-012:1272010:1272998 [5] NCCL INFO Channel 07/0 : 5[5] -> 13[5] [send] via NET/IB/6/GDRDMA cnode7-012:1272010:1272998 [5] NCCL INFO Channel 08/0 : 5[5] -> 13[5] [send] via NET/IB/6/GDRDMA cnode7-012:1272010:1272998 [5] NCCL INFO Channel 09/0 : 5[5] -> 13[5] [send] via NET/IB/6/GDRDMA cnode7-012:1272010:1272998 [5] NCCL INFO Channel 10/0 : 5[5] -> 13[5] [send] via NET/IB/6/GDRDMA cnode7-012:1272010:1272998 [5] NCCL INFO Channel 11/0 : 5[5] -> 13[5] [send] via NET/IB/6/GDRDMA cnode7-012:1272010:1272998 [5] NCCL INFO Channel 14/0 : 5[5] -> 13[5] [send] via NET/IB/6/GDRDMA cnode7-012:1272010:1272998 [5] NCCL INFO Channel 15/0 : 5[5] -> 13[5] [send] via NET/IB/6/GDRDMA cnode7-013:736484:737409 [3] NCCL INFO Channel 12/0 : 3[3] -> 11[3] [receive] via NET/IB/3/GDRDMA cnode7-013:736484:737409 [3] NCCL INFO Channel 13/0 : 3[3] -> 11[3] [receive] via NET/IB/3/GDRDMA cnode7-013:736484:737409 [3] NCCL INFO Channel 14/0 : 3[3] -> 11[3] [receive] via NET/IB/3/GDRDMA cnode7-013:736484:737409 [3] NCCL INFO Channel 15/0 : 3[3] -> 11[3] [receive] via NET/IB/3/GDRDMA cnode7-013:736484:737409 [3] NCCL INFO Channel 00/0 : 11[3] -> 3[3] [send] via NET/IB/3/GDRDMA cnode7-013:736484:737409 [3] NCCL INFO Channel 01/0 : 11[3] -> 3[3] [send] via NET/IB/3/GDRDMA cnode7-013:736484:737409 [3] NCCL INFO Channel 02/0 : 11[3] -> 3[3] [send] via NET/IB/3/GDRDMA cnode7-013:736484:737409 [3] NCCL INFO Channel 04/0 : 11[3] -> 3[3] [send] via NET/IB/3/GDRDMA cnode7-013:736484:737409 [3] NCCL INFO Channel 05/0 : 11[3] -> 3[3] [send] via NET/IB/3/GDRDMA cnode7-013:736484:737409 [3] NCCL INFO Channel 06/0 : 11[3] -> 3[3] [send] via NET/IB/3/GDRDMA cnode7-013:736484:737409 [3] NCCL INFO Channel 07/0 : 11[3] -> 3[3] [send] via NET/IB/3/GDRDMA cnode7-013:736484:737409 [3] NCCL INFO Channel 08/0 : 11[3] -> 3[3] [send] via NET/IB/3/GDRDMA cnode7-013:736484:737409 [3] NCCL INFO Channel 09/0 : 11[3] -> 3[3] [send] via NET/IB/3/GDRDMA cnode7-013:736484:737409 [3] NCCL INFO Channel 10/0 : 11[3] -> 3[3] [send] via NET/IB/3/GDRDMA cnode7-013:736484:737409 [3] NCCL INFO Channel 12/0 : 11[3] -> 3[3] [send] via NET/IB/3/GDRDMA cnode7-013:736484:737409 [3] NCCL INFO Channel 13/0 : 11[3] -> 3[3] [send] via NET/IB/3/GDRDMA cnode7-013:736484:737409 [3] NCCL INFO Channel 14/0 : 11[3] -> 3[3] [send] via NET/IB/3/GDRDMA cnode7-013:736484:737409 [3] NCCL INFO Channel 15/0 : 11[3] -> 3[3] [send] via NET/IB/3/GDRDMA cnode7-013:736488:737372 [7] NCCL INFO Channel 10/0 : 7[7] -> 15[7] [receive] via NET/IB/8/GDRDMA cnode7-013:736488:737372 [7] NCCL INFO Channel 11/0 : 7[7] -> 15[7] [receive] via NET/IB/8/GDRDMA cnode7-013:736488:737372 [7] NCCL INFO Channel 12/0 : 7[7] -> 15[7] [receive] via NET/IB/8/GDRDMA cnode7-013:736488:737372 [7] NCCL INFO Channel 13/0 : 7[7] -> 15[7] [receive] via NET/IB/8/GDRDMA cnode7-013:736488:737372 [7] NCCL INFO Channel 00/0 : 15[7] -> 7[7] [send] via NET/IB/8/GDRDMA cnode7-013:736488:737372 [7] NCCL INFO Channel 01/0 : 15[7] -> 7[7] [send] via NET/IB/8/GDRDMA cnode7-013:736488:737372 [7] NCCL INFO Channel 02/0 : 15[7] -> 7[7] [send] via NET/IB/8/GDRDMA cnode7-013:736488:737372 [7] NCCL INFO Channel 03/0 : 15[7] -> 7[7] [send] via NET/IB/8/GDRDMA cnode7-013:736488:737372 [7] NCCL INFO Channel 04/0 : 15[7] -> 7[7] [send] via NET/IB/8/GDRDMA cnode7-013:736488:737372 [7] NCCL INFO Channel 05/0 : 15[7] -> 7[7] [send] via NET/IB/8/GDRDMA cnode7-013:736488:737372 [7] NCCL INFO Channel 06/0 : 15[7] -> 7[7] [send] via NET/IB/8/GDRDMA cnode7-013:736488:737372 [7] NCCL INFO Channel 08/0 : 15[7] -> 7[7] [send] via NET/IB/8/GDRDMA cnode7-013:736488:737372 [7] NCCL INFO Channel 09/0 : 15[7] -> 7[7] [send] via NET/IB/8/GDRDMA cnode7-013:736488:737372 [7] NCCL INFO Channel 10/0 : 15[7] -> 7[7] [send] via NET/IB/8/GDRDMA cnode7-013:736488:737372 [7] NCCL INFO Channel 11/0 : 15[7] -> 7[7] [send] via NET/IB/8/GDRDMA cnode7-013:736488:737372 [7] NCCL INFO Channel 12/0 : 15[7] -> 7[7] [send] via NET/IB/8/GDRDMA cnode7-013:736488:737372 [7] NCCL INFO Channel 13/0 : 15[7] -> 7[7] [send] via NET/IB/8/GDRDMA cnode7-013:736488:737372 [7] NCCL INFO Channel 14/0 : 15[7] -> 7[7] [send] via NET/IB/8/GDRDMA cnode7-013:736487:737367 [6] NCCL INFO Channel 10/0 : 6[6] -> 14[6] [receive] via NET/IB/7/GDRDMA cnode7-013:736487:737367 [6] NCCL INFO Channel 11/0 : 6[6] -> 14[6] [receive] via NET/IB/7/GDRDMA cnode7-013:736487:737367 [6] NCCL INFO Channel 12/0 : 6[6] -> 14[6] [receive] via NET/IB/7/GDRDMA cnode7-013:736487:737367 [6] NCCL INFO Channel 13/0 : 6[6] -> 14[6] [receive] via NET/IB/7/GDRDMA cnode7-013:736487:737367 [6] NCCL INFO Channel 00/0 : 14[6] -> 6[6] [send] via NET/IB/7/GDRDMA cnode7-013:736487:737367 [6] NCCL INFO Channel 01/0 : 14[6] -> 6[6] [send] via NET/IB/7/GDRDMA cnode7-013:736487:737367 [6] NCCL INFO Channel 02/0 : 14[6] -> 6[6] [send] via NET/IB/7/GDRDMA cnode7-013:736487:737367 [6] NCCL INFO Channel 03/0 : 14[6] -> 6[6] [send] via NET/IB/7/GDRDMA cnode7-013:736487:737367 [6] NCCL INFO Channel 04/0 : 14[6] -> 6[6] [send] via NET/IB/7/GDRDMA cnode7-013:736487:737367 [6] NCCL INFO Channel 05/0 : 14[6] -> 6[6] [send] via NET/IB/7/GDRDMA cnode7-013:736487:737367 [6] NCCL INFO Channel 07/0 : 14[6] -> 6[6] [send] via NET/IB/7/GDRDMA cnode7-013:736487:737367 [6] NCCL INFO Channel 08/0 : 14[6] -> 6[6] [send] via NET/IB/7/GDRDMA cnode7-013:736487:737367 [6] NCCL INFO Channel 09/0 : 14[6] -> 6[6] [send] via NET/IB/7/GDRDMA cnode7-013:736487:737367 [6] NCCL INFO Channel 10/0 : 14[6] -> 6[6] [send] via NET/IB/7/GDRDMA cnode7-013:736487:737367 [6] NCCL INFO Channel 11/0 : 14[6] -> 6[6] [send] via NET/IB/7/GDRDMA cnode7-013:736487:737367 [6] NCCL INFO Channel 12/0 : 14[6] -> 6[6] [send] via NET/IB/7/GDRDMA cnode7-013:736487:737367 [6] NCCL INFO Channel 13/0 : 14[6] -> 6[6] [send] via NET/IB/7/GDRDMA cnode7-013:736487:737367 [6] NCCL INFO Channel 15/0 : 14[6] -> 6[6] [send] via NET/IB/7/GDRDMA cnode7-012:1272008:1272929 [3] NCCL INFO Channel 09/0 : 11[3] -> 3[3] [receive] via NET/IB/3/GDRDMA cnode7-012:1272008:1272929 [3] NCCL INFO Channel 10/0 : 11[3] -> 3[3] [receive] via NET/IB/3/GDRDMA cnode7-012:1272008:1272929 [3] NCCL INFO Channel 12/0 : 11[3] -> 3[3] [receive] via NET/IB/3/GDRDMA cnode7-012:1272008:1272929 [3] NCCL INFO Channel 13/0 : 11[3] -> 3[3] [receive] via NET/IB/3/GDRDMA cnode7-012:1272008:1272929 [3] NCCL INFO Channel 14/0 : 11[3] -> 3[3] [receive] via NET/IB/3/GDRDMA cnode7-012:1272008:1272929 [3] NCCL INFO Channel 15/0 : 11[3] -> 3[3] [receive] via NET/IB/3/GDRDMA cnode7-012:1272008:1272929 [3] NCCL INFO Channel 00/0 : 3[3] -> 11[3] [send] via NET/IB/3/GDRDMA cnode7-012:1272008:1272929 [3] NCCL INFO Channel 01/0 : 3[3] -> 11[3] [send] via NET/IB/3/GDRDMA cnode7-012:1272008:1272929 [3] NCCL INFO Channel 04/0 : 3[3] -> 11[3] [send] via NET/IB/3/GDRDMA cnode7-012:1272008:1272929 [3] NCCL INFO Channel 05/0 : 3[3] -> 11[3] [send] via NET/IB/3/GDRDMA cnode7-012:1272008:1272929 [3] NCCL INFO Channel 06/0 : 3[3] -> 11[3] [send] via NET/IB/3/GDRDMA cnode7-012:1272008:1272929 [3] NCCL INFO Channel 07/0 : 3[3] -> 11[3] [send] via NET/IB/3/GDRDMA cnode7-012:1272008:1272929 [3] NCCL INFO Channel 08/0 : 3[3] -> 11[3] [send] via NET/IB/3/GDRDMA cnode7-012:1272008:1272929 [3] NCCL INFO Channel 09/0 : 3[3] -> 11[3] [send] via NET/IB/3/GDRDMA cnode7-012:1272008:1272929 [3] NCCL INFO Channel 12/0 : 3[3] -> 11[3] [send] via NET/IB/3/GDRDMA cnode7-012:1272008:1272929 [3] NCCL INFO Channel 13/0 : 3[3] -> 11[3] [send] via NET/IB/3/GDRDMA cnode7-012:1272008:1272929 [3] NCCL INFO Channel 14/0 : 3[3] -> 11[3] [send] via NET/IB/3/GDRDMA cnode7-012:1272008:1272929 [3] NCCL INFO Channel 15/0 : 3[3] -> 11[3] [send] via NET/IB/3/GDRDMA cnode7-013:736482:737436 [1] NCCL INFO Channel 12/0 : 1[1] -> 9[1] [receive] via NET/IB/1/GDRDMA cnode7-013:736482:737436 [1] NCCL INFO Channel 13/0 : 1[1] -> 9[1] [receive] via NET/IB/1/GDRDMA cnode7-013:736482:737436 [1] NCCL INFO Channel 14/0 : 1[1] -> 9[1] [receive] via NET/IB/1/GDRDMA cnode7-013:736482:737436 [1] NCCL INFO Channel 15/0 : 1[1] -> 9[1] [receive] via NET/IB/1/GDRDMA cnode7-013:736482:737436 [1] NCCL INFO Channel 00/0 : 9[1] -> 1[1] [send] via NET/IB/1/GDRDMA cnode7-013:736482:737436 [1] NCCL INFO Channel 02/0 : 9[1] -> 1[1] [send] via NET/IB/1/GDRDMA cnode7-013:736482:737436 [1] NCCL INFO Channel 03/0 : 9[1] -> 1[1] [send] via NET/IB/1/GDRDMA cnode7-013:736482:737436 [1] NCCL INFO Channel 04/0 : 9[1] -> 1[1] [send] via NET/IB/1/GDRDMA cnode7-013:736482:737436 [1] NCCL INFO Channel 05/0 : 9[1] -> 1[1] [send] via NET/IB/1/GDRDMA cnode7-013:736482:737436 [1] NCCL INFO Channel 06/0 : 9[1] -> 1[1] [send] via NET/IB/1/GDRDMA cnode7-013:736482:737436 [1] NCCL INFO Channel 07/0 : 9[1] -> 1[1] [send] via NET/IB/1/GDRDMA cnode7-013:736482:737436 [1] NCCL INFO Channel 08/0 : 9[1] -> 1[1] [send] via NET/IB/1/GDRDMA cnode7-013:736482:737436 [1] NCCL INFO Channel 10/0 : 9[1] -> 1[1] [send] via NET/IB/1/GDRDMA cnode7-013:736482:737436 [1] NCCL INFO Channel 11/0 : 9[1] -> 1[1] [send] via NET/IB/1/GDRDMA cnode7-013:736482:737436 [1] NCCL INFO Channel 12/0 : 9[1] -> 1[1] [send] via NET/IB/1/GDRDMA cnode7-013:736482:737436 [1] NCCL INFO Channel 13/0 : 9[1] -> 1[1] [send] via NET/IB/1/GDRDMA cnode7-013:736482:737436 [1] NCCL INFO Channel 14/0 : 9[1] -> 1[1] [send] via NET/IB/1/GDRDMA cnode7-013:736482:737436 [1] NCCL INFO Channel 15/0 : 9[1] -> 1[1] [send] via NET/IB/1/GDRDMA cnode7-012:1272007:1272925 [2] NCCL INFO Channel 09/0 : 10[2] -> 2[2] [receive] via NET/IB/2/GDRDMA cnode7-012:1272007:1272925 [2] NCCL INFO Channel 11/0 : 10[2] -> 2[2] [receive] via NET/IB/2/GDRDMA cnode7-012:1272007:1272925 [2] NCCL INFO Channel 12/0 : 10[2] -> 2[2] [receive] via NET/IB/2/GDRDMA cnode7-012:1272007:1272925 [2] NCCL INFO Channel 13/0 : 10[2] -> 2[2] [receive] via NET/IB/2/GDRDMA cnode7-012:1272007:1272925 [2] NCCL INFO Channel 14/0 : 10[2] -> 2[2] [receive] via NET/IB/2/GDRDMA cnode7-012:1272007:1272925 [2] NCCL INFO Channel 15/0 : 10[2] -> 2[2] [receive] via NET/IB/2/GDRDMA cnode7-012:1272007:1272925 [2] NCCL INFO Channel 00/0 : 2[2] -> 10[2] [send] via NET/IB/2/GDRDMA cnode7-012:1272007:1272925 [2] NCCL INFO Channel 01/0 : 2[2] -> 10[2] [send] via NET/IB/2/GDRDMA cnode7-012:1272007:1272925 [2] NCCL INFO Channel 04/0 : 2[2] -> 10[2] [send] via NET/IB/2/GDRDMA cnode7-012:1272007:1272925 [2] NCCL INFO Channel 05/0 : 2[2] -> 10[2] [send] via NET/IB/2/GDRDMA cnode7-012:1272007:1272925 [2] NCCL INFO Channel 06/0 : 2[2] -> 10[2] [send] via NET/IB/2/GDRDMA cnode7-012:1272007:1272925 [2] NCCL INFO Channel 07/0 : 2[2] -> 10[2] [send] via NET/IB/2/GDRDMA cnode7-012:1272007:1272925 [2] NCCL INFO Channel 08/0 : 2[2] -> 10[2] [send] via NET/IB/2/GDRDMA cnode7-012:1272007:1272925 [2] NCCL INFO Channel 09/0 : 2[2] -> 10[2] [send] via NET/IB/2/GDRDMA cnode7-012:1272007:1272925 [2] NCCL INFO Channel 12/0 : 2[2] -> 10[2] [send] via NET/IB/2/GDRDMA cnode7-012:1272007:1272925 [2] NCCL INFO Channel 13/0 : 2[2] -> 10[2] [send] via NET/IB/2/GDRDMA cnode7-012:1272007:1272925 [2] NCCL INFO Channel 14/0 : 2[2] -> 10[2] [send] via NET/IB/2/GDRDMA cnode7-012:1272007:1272925 [2] NCCL INFO Channel 15/0 : 2[2] -> 10[2] [send] via NET/IB/2/GDRDMA cnode7-013:736485:737373 [4] NCCL INFO Channel 10/0 : 4[4] -> 12[4] [receive] via NET/IB/4/GDRDMA cnode7-013:736485:737373 [4] NCCL INFO Channel 11/0 : 4[4] -> 12[4] [receive] via NET/IB/4/GDRDMA cnode7-013:736485:737373 [4] NCCL INFO Channel 14/0 : 4[4] -> 12[4] [receive] via NET/IB/4/GDRDMA cnode7-013:736485:737373 [4] NCCL INFO Channel 15/0 : 4[4] -> 12[4] [receive] via NET/IB/4/GDRDMA cnode7-013:736485:737373 [4] NCCL INFO Channel 00/0 : 12[4] -> 4[4] [send] via NET/IB/4/GDRDMA cnode7-013:736485:737373 [4] NCCL INFO Channel 01/0 : 12[4] -> 4[4] [send] via NET/IB/4/GDRDMA cnode7-013:736485:737373 [4] NCCL INFO Channel 02/0 : 12[4] -> 4[4] [send] via NET/IB/4/GDRDMA cnode7-013:736485:737373 [4] NCCL INFO Channel 03/0 : 12[4] -> 4[4] [send] via NET/IB/4/GDRDMA cnode7-013:736485:737373 [4] NCCL INFO Channel 05/0 : 12[4] -> 4[4] [send] via NET/IB/4/GDRDMA cnode7-013:736485:737373 [4] NCCL INFO Channel 06/0 : 12[4] -> 4[4] [send] via NET/IB/4/GDRDMA cnode7-013:736485:737373 [4] NCCL INFO Channel 07/0 : 12[4] -> 4[4] [send] via NET/IB/4/GDRDMA cnode7-013:736485:737373 [4] NCCL INFO Channel 08/0 : 12[4] -> 4[4] [send] via NET/IB/4/GDRDMA cnode7-013:736485:737373 [4] NCCL INFO Channel 09/0 : 12[4] -> 4[4] [send] via NET/IB/4/GDRDMA cnode7-013:736485:737373 [4] NCCL INFO Channel 10/0 : 12[4] -> 4[4] [send] via NET/IB/4/GDRDMA cnode7-013:736485:737373 [4] NCCL INFO Channel 11/0 : 12[4] -> 4[4] [send] via NET/IB/4/GDRDMA cnode7-013:736485:737373 [4] NCCL INFO Channel 13/0 : 12[4] -> 4[4] [send] via NET/IB/4/GDRDMA cnode7-013:736485:737373 [4] NCCL INFO Channel 14/0 : 12[4] -> 4[4] [send] via NET/IB/4/GDRDMA cnode7-013:736485:737373 [4] NCCL INFO Channel 15/0 : 12[4] -> 4[4] [send] via NET/IB/4/GDRDMA cnode7-013:736486:737423 [5] NCCL INFO Channel 10/0 : 5[5] -> 13[5] [receive] via NET/IB/6/GDRDMA cnode7-013:736486:737423 [5] NCCL INFO Channel 11/0 : 5[5] -> 13[5] [receive] via NET/IB/6/GDRDMA cnode7-013:736486:737423 [5] NCCL INFO Channel 14/0 : 5[5] -> 13[5] [receive] via NET/IB/6/GDRDMA cnode7-013:736486:737423 [5] NCCL INFO Channel 15/0 : 5[5] -> 13[5] [receive] via NET/IB/6/GDRDMA cnode7-013:736486:737423 [5] NCCL INFO Channel 00/0 : 13[5] -> 5[5] [send] via NET/IB/6/GDRDMA cnode7-013:736486:737423 [5] NCCL INFO Channel 01/0 : 13[5] -> 5[5] [send] via NET/IB/6/GDRDMA cnode7-013:736486:737423 [5] NCCL INFO Channel 02/0 : 13[5] -> 5[5] [send] via NET/IB/6/GDRDMA cnode7-013:736486:737423 [5] NCCL INFO Channel 03/0 : 13[5] -> 5[5] [send] via NET/IB/6/GDRDMA cnode7-013:736486:737423 [5] NCCL INFO Channel 04/0 : 13[5] -> 5[5] [send] via NET/IB/6/GDRDMA cnode7-013:736486:737423 [5] NCCL INFO Channel 06/0 : 13[5] -> 5[5] [send] via NET/IB/6/GDRDMA cnode7-013:736486:737423 [5] NCCL INFO Channel 07/0 : 13[5] -> 5[5] [send] via NET/IB/6/GDRDMA cnode7-013:736486:737423 [5] NCCL INFO Channel 08/0 : 13[5] -> 5[5] [send] via NET/IB/6/GDRDMA cnode7-013:736486:737423 [5] NCCL INFO Channel 09/0 : 13[5] -> 5[5] [send] via NET/IB/6/GDRDMA cnode7-013:736486:737423 [5] NCCL INFO Channel 10/0 : 13[5] -> 5[5] [send] via NET/IB/6/GDRDMA cnode7-013:736486:737423 [5] NCCL INFO Channel 11/0 : 13[5] -> 5[5] [send] via NET/IB/6/GDRDMA cnode7-013:736486:737423 [5] NCCL INFO Channel 12/0 : 13[5] -> 5[5] [send] via NET/IB/6/GDRDMA cnode7-013:736486:737423 [5] NCCL INFO Channel 14/0 : 13[5] -> 5[5] [send] via NET/IB/6/GDRDMA cnode7-013:736486:737423 [5] NCCL INFO Channel 15/0 : 13[5] -> 5[5] [send] via NET/IB/6/GDRDMA cnode7-012:1272012:1272930 [7] NCCL INFO Channel 04/0 : 7[7] -> 15[7] [send] via NET/IB/8/GDRDMA cnode7-012:1272012:1272930 [7] NCCL INFO Channel 05/0 : 7[7] -> 15[7] [send] via NET/IB/8/GDRDMA cnode7-012:1272012:1272930 [7] NCCL INFO Channel 08/0 : 7[7] -> 15[7] [send] via NET/IB/8/GDRDMA cnode7-012:1272012:1272930 [7] NCCL INFO Channel 09/0 : 7[7] -> 15[7] [send] via NET/IB/8/GDRDMA cnode7-012:1272012:1272930 [7] NCCL INFO Channel 10/0 : 7[7] -> 15[7] [send] via NET/IB/8/GDRDMA cnode7-012:1272012:1272930 [7] NCCL INFO Channel 11/0 : 7[7] -> 15[7] [send] via NET/IB/8/GDRDMA cnode7-012:1272012:1272930 [7] NCCL INFO Channel 12/0 : 7[7] -> 15[7] [send] via NET/IB/8/GDRDMA cnode7-012:1272012:1272930 [7] NCCL INFO Channel 13/0 : 7[7] -> 15[7] [send] via NET/IB/8/GDRDMA cnode7-013:736481:737376 [0] NCCL INFO Channel 13/0 : 0[0] -> 8[0] [receive] via NET/IB/0/GDRDMA cnode7-013:736481:737376 [0] NCCL INFO Channel 14/0 : 0[0] -> 8[0] [receive] via NET/IB/0/GDRDMA cnode7-013:736481:737376 [0] NCCL INFO Channel 15/0 : 0[0] -> 8[0] [receive] via NET/IB/0/GDRDMA cnode7-013:736481:737376 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [send] via NET/IB/0/GDRDMA cnode7-013:736481:737376 [0] NCCL INFO Channel 02/0 : 8[0] -> 0[0] [send] via NET/IB/0/GDRDMA cnode7-013:736481:737376 [0] NCCL INFO Channel 03/0 : 8[0] -> 0[0] [send] via NET/IB/0/GDRDMA cnode7-013:736481:737376 [0] NCCL INFO Channel 04/0 : 8[0] -> 0[0] [send] via NET/IB/0/GDRDMA cnode7-013:736481:737376 [0] NCCL INFO Channel 05/0 : 8[0] -> 0[0] [send] via NET/IB/0/GDRDMA cnode7-013:736481:737376 [0] NCCL INFO Channel 06/0 : 8[0] -> 0[0] [send] via NET/IB/0/GDRDMA cnode7-013:736481:737376 [0] NCCL INFO Channel 07/0 : 8[0] -> 0[0] [send] via NET/IB/0/GDRDMA cnode7-013:736481:737376 [0] NCCL INFO Channel 09/0 : 8[0] -> 0[0] [send] via NET/IB/0/GDRDMA cnode7-013:736481:737376 [0] NCCL INFO Channel 10/0 : 8[0] -> 0[0] [send] via NET/IB/0/GDRDMA cnode7-013:736481:737376 [0] NCCL INFO Channel 11/0 : 8[0] -> 0[0] [send] via NET/IB/0/GDRDMA cnode7-013:736481:737376 [0] NCCL INFO Channel 12/0 : 8[0] -> 0[0] [send] via NET/IB/0/GDRDMA cnode7-013:736481:737376 [0] NCCL INFO Channel 13/0 : 8[0] -> 0[0] [send] via NET/IB/0/GDRDMA cnode7-013:736481:737376 [0] NCCL INFO Channel 14/0 : 8[0] -> 0[0] [send] via NET/IB/0/GDRDMA cnode7-013:736481:737376 [0] NCCL INFO Channel 15/0 : 8[0] -> 0[0] [send] via NET/IB/0/GDRDMA cnode7-012:1272010:1272998 [5] NCCL INFO Connected NVLS tree cnode7-012:1272010:1272998 [5] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512 cnode7-012:1272010:1272998 [5] NCCL INFO 16 coll channels, 0 collnet channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer cnode7-013:736484:737409 [3] NCCL INFO Connected NVLS tree cnode7-013:736484:737409 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512 cnode7-013:736484:737409 [3] NCCL INFO 16 coll channels, 0 collnet channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer cnode7-013:736483:737422 [2] NCCL INFO Connected NVLS tree cnode7-013:736483:737422 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512 cnode7-013:736483:737422 [2] NCCL INFO 16 coll channels, 0 collnet channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer cnode7-013:736482:737436 [1] NCCL INFO Connected NVLS tree cnode7-013:736482:737436 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512 cnode7-013:736482:737436 [1] NCCL INFO 16 coll channels, 0 collnet channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer cnode7-013:736488:737372 [7] NCCL INFO Connected NVLS tree cnode7-013:736488:737372 [7] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512 cnode7-013:736488:737372 [7] NCCL INFO 16 coll channels, 0 collnet channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer cnode7-013:736487:737367 [6] NCCL INFO Connected NVLS tree cnode7-013:736487:737367 [6] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512 cnode7-013:736487:737367 [6] NCCL INFO 16 coll channels, 0 collnet channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer cnode7-013:736486:737423 [5] NCCL INFO Connected NVLS tree cnode7-013:736486:737423 [5] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512 cnode7-013:736486:737423 [5] NCCL INFO 16 coll channels, 0 collnet channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer cnode7-012:1272008:1272929 [3] NCCL INFO Connected NVLS tree cnode7-012:1272008:1272929 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512 cnode7-012:1272008:1272929 [3] NCCL INFO 16 coll channels, 0 collnet channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer cnode7-013:736481:737376 [0] NCCL INFO Connected NVLS tree cnode7-013:736481:737376 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512 cnode7-013:736481:737376 [0] NCCL INFO 16 coll channels, 0 collnet channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer cnode7-012:1272007:1272925 [2] NCCL INFO Connected NVLS tree cnode7-012:1272007:1272925 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512 cnode7-012:1272007:1272925 [2] NCCL INFO 16 coll channels, 0 collnet channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer cnode7-013:736485:737373 [4] NCCL INFO Connected NVLS tree cnode7-013:736485:737373 [4] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512 cnode7-013:736485:737373 [4] NCCL INFO 16 coll channels, 0 collnet channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer cnode7-013:736486:737423 [5] NCCL INFO comm 0x555581618ce0 rank 13 nranks 16 cudaDev 5 nvmlDev 5 busId c3000 commId 0x4e42d6174620c76 - Init COMPLETE cnode7-012:1272006:1272926 [1] NCCL INFO Connected NVLS tree cnode7-012:1272006:1272926 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512 cnode7-012:1272006:1272926 [1] NCCL INFO 16 coll channels, 0 collnet channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer cnode7-012:1272012:1272930 [7] NCCL INFO Connected NVLS tree cnode7-012:1272012:1272930 [7] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512 cnode7-012:1272012:1272930 [7] NCCL INFO 16 coll channels, 0 collnet channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer cnode7-013:736483:737422 [2] NCCL INFO comm 0x55558161e2b0 rank 10 nranks 16 cudaDev 2 nvmlDev 2 busId 52000 commId 0x4e42d6174620c76 - Init COMPLETE cnode7-013:736485:737373 [4] NCCL INFO comm 0x555560c19b70 rank 12 nranks 16 cudaDev 4 nvmlDev 4 busId 9d000 commId 0x4e42d6174620c76 - Init COMPLETE cnode7-013:736488:737372 [7] NCCL INFO comm 0x555560c15d00 rank 15 nranks 16 cudaDev 7 nvmlDev 7 busId df000 commId 0x4e42d6174620c76 - Init COMPLETE cnode7-013:736482:737436 [1] NCCL INFO comm 0x555580e21bb0 rank 9 nranks 16 cudaDev 1 nvmlDev 1 busId 43000 commId 0x4e42d6174620c76 - Init COMPLETE cnode7-013:736484:737409 [3] NCCL INFO comm 0x55557b419c80 rank 11 nranks 16 cudaDev 3 nvmlDev 3 busId 61000 commId 0x4e42d6174620c76 - Init COMPLETE cnode7-013:736481:737376 [0] NCCL INFO comm 0x555563b6a610 rank 8 nranks 16 cudaDev 0 nvmlDev 0 busId 1b000 commId 0x4e42d6174620c76 - Init COMPLETE cnode7-012:1272005:1272924 [0] NCCL INFO Connected NVLS tree cnode7-012:1272005:1272924 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512 cnode7-012:1272005:1272924 [0] NCCL INFO 16 coll channels, 0 collnet channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer cnode7-012:1272011:1272928 [6] NCCL INFO Connected NVLS tree cnode7-012:1272011:1272928 [6] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512 cnode7-012:1272011:1272928 [6] NCCL INFO 16 coll channels, 0 collnet channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer cnode7-012:1272009:1272927 [4] NCCL INFO Connected NVLS tree cnode7-012:1272009:1272927 [4] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512 cnode7-012:1272009:1272927 [4] NCCL INFO 16 coll channels, 0 collnet channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer cnode7-013:736487:737367 [6] NCCL INFO comm 0x555561a1a9e0 rank 14 nranks 16 cudaDev 6 nvmlDev 6 busId d1000 commId 0x4e42d6174620c76 - Init COMPLETE cnode7-012:1272009:1272927 [4] NCCL INFO comm 0x55558161a360 rank 4 nranks 16 cudaDev 4 nvmlDev 4 busId 9d000 commId 0x4e42d6174620c76 - Init COMPLETE cnode7-012:1272011:1272928 [6] NCCL INFO comm 0x55557c81dd70 rank 6 nranks 16 cudaDev 6 nvmlDev 6 busId d1000 commId 0x4e42d6174620c76 - Init COMPLETE cnode7-012:1272007:1272925 [2] NCCL INFO comm 0x55558061a960 rank 2 nranks 16 cudaDev 2 nvmlDev 2 busId 52000 commId 0x4e42d6174620c76 - Init COMPLETE cnode7-012:1272012:1272930 [7] NCCL INFO comm 0x55557b820a90 rank 7 nranks 16 cudaDev 7 nvmlDev 7 busId df000 commId 0x4e42d6174620c76 - Init COMPLETE cnode7-012:1272010:1272998 [5] NCCL INFO comm 0x555572620330 rank 5 nranks 16 cudaDev 5 nvmlDev 5 busId c3000 commId 0x4e42d6174620c76 - Init COMPLETE cnode7-012:1272008:1272929 [3] NCCL INFO comm 0x55557b81cee0 rank 3 nranks 16 cudaDev 3 nvmlDev 3 busId 61000 commId 0x4e42d6174620c76 - Init COMPLETE cnode7-012:1272006:1272926 [1] NCCL INFO comm 0x55558161ded0 rank 1 nranks 16 cudaDev 1 nvmlDev 1 busId 43000 commId 0x4e42d6174620c76 - Init COMPLETE cnode7-012:1272005:1272924 [0] NCCL INFO comm 0x555563bd8350 rank 0 nranks 16 cudaDev 0 nvmlDev 0 busId 1b000 commId 0x4e42d6174620c76 - Init COMPLETE [2024-12-05 16:34:12,460] [INFO] [logging.py:129:log_dist] [Rank 0] DeepSpeed Flops Profiler Enabled: False [2024-12-05 16:34:12,469] [INFO] [logging.py:129:log_dist] [Rank 0] Using DeepSpeed Optimizer param name adam as basic optimizer [2024-12-05 16:34:12,470] [INFO] [logging.py:129:log_dist] [Rank 0] Removing param_group that has no 'params' in the basic Optimizer [2024-12-05 16:34:12,502] [INFO] [logging.py:129:log_dist] [Rank 0] DeepSpeed Basic Optimizer = FusedAdam [2024-12-05 16:34:12,502] [INFO] [utils.py:59:is_zero_supported_optimizer] Checking ZeRO support for optimizer=FusedAdam type= [2024-12-05 16:34:12,502] [INFO] [logging.py:129:log_dist] [Rank 0] Creating torch.bfloat16 ZeRO stage 2 optimizer [2024-12-05 16:34:12,502] [INFO] [stage_1_and_2.py:149:__init__] Reduce bucket size 500000000 [2024-12-05 16:34:12,502] [INFO] [stage_1_and_2.py:150:__init__] Allgather bucket size 500000000 [2024-12-05 16:34:12,502] [INFO] [stage_1_and_2.py:151:__init__] CPU Offload: False [2024-12-05 16:34:12,502] [INFO] [stage_1_and_2.py:152:__init__] Round robin gradient partitioning: False [2024-12-05 16:34:14,018] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt... [2024-12-05 16:34:14,018] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt... /mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/site-packages/deepspeed/runtime/checkpoint_engine/torch_checkpoint_engine.py:28: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature. partition = torch.load(path, map_location=map_location) /mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/site-packages/deepspeed/runtime/checkpoint_engine/torch_checkpoint_engine.py:28: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature. partition = torch.load(path, map_location=map_location) [2024-12-05 16:34:14,021] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt... /mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/site-packages/deepspeed/runtime/checkpoint_engine/torch_checkpoint_engine.py:28: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature. partition = torch.load(path, map_location=map_location) [2024-12-05 16:34:14,074] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt... /mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/site-packages/deepspeed/runtime/checkpoint_engine/torch_checkpoint_engine.py:28: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature. partition = torch.load(path, map_location=map_location) [2024-12-05 16:34:14,083] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt... /mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/site-packages/deepspeed/runtime/checkpoint_engine/torch_checkpoint_engine.py:28: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature. partition = torch.load(path, map_location=map_location) [2024-12-05 16:34:14,084] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt... /mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/site-packages/deepspeed/runtime/checkpoint_engine/torch_checkpoint_engine.py:28: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature. partition = torch.load(path, map_location=map_location) [2024-12-05 16:34:14,238] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt... [2024-12-05 16:34:14,238] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt... [2024-12-05 16:34:14,238] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt... /mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/site-packages/deepspeed/runtime/checkpoint_engine/torch_checkpoint_engine.py:28: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature. partition = torch.load(path, map_location=map_location) /mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/site-packages/deepspeed/runtime/checkpoint_engine/torch_checkpoint_engine.py:28: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature. partition = torch.load(path, map_location=map_location) /mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/site-packages/deepspeed/runtime/checkpoint_engine/torch_checkpoint_engine.py:28: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature. partition = torch.load(path, map_location=map_location) [2024-12-05 16:34:14,268] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt... /mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/site-packages/deepspeed/runtime/checkpoint_engine/torch_checkpoint_engine.py:28: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature. partition = torch.load(path, map_location=map_location) [2024-12-05 16:34:14,277] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt... /mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/site-packages/deepspeed/runtime/checkpoint_engine/torch_checkpoint_engine.py:28: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature. partition = torch.load(path, map_location=map_location) [2024-12-05 16:34:14,320] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt... /mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/site-packages/deepspeed/runtime/checkpoint_engine/torch_checkpoint_engine.py:28: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature. partition = torch.load(path, map_location=map_location) [2024-12-05 16:34:14,187] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt... /mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/site-packages/deepspeed/runtime/checkpoint_engine/torch_checkpoint_engine.py:28: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature. partition = torch.load(path, map_location=map_location) [2024-12-05 16:34:14,252] [INFO] [utils.py:781:see_memory_usage] Before initializing optimizer states [2024-12-05 16:34:14,255] [INFO] [utils.py:782:see_memory_usage] MA 1.37 GB Max_MA 1.45 GB CA 1.45 GB Max_CA 1 GB [2024-12-05 16:34:14,255] [INFO] [utils.py:789:see_memory_usage] CPU Virtual Memory: used = 74.57 GB, percent = 3.7% [2024-12-05 16:34:14,571] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt... /mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/site-packages/deepspeed/runtime/checkpoint_engine/torch_checkpoint_engine.py:28: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature. partition = torch.load(path, map_location=map_location) [2024-12-05 16:34:14,453] [INFO] [utils.py:781:see_memory_usage] After initializing optimizer states [2024-12-05 16:34:14,454] [INFO] [utils.py:782:see_memory_usage] MA 1.37 GB Max_MA 1.52 GB CA 1.6 GB Max_CA 2 GB [2024-12-05 16:34:14,454] [INFO] [utils.py:789:see_memory_usage] CPU Virtual Memory: used = 76.93 GB, percent = 3.8% [2024-12-05 16:34:14,454] [INFO] [stage_1_and_2.py:544:__init__] optimizer state initialized [2024-12-05 16:34:14,723] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt... /mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/site-packages/deepspeed/runtime/checkpoint_engine/torch_checkpoint_engine.py:28: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature. partition = torch.load(path, map_location=map_location) [2024-12-05 16:34:14,620] [INFO] [utils.py:781:see_memory_usage] After initializing ZeRO optimizer [2024-12-05 16:34:14,621] [INFO] [utils.py:782:see_memory_usage] MA 1.37 GB Max_MA 1.37 GB CA 1.6 GB Max_CA 2 GB [2024-12-05 16:34:14,621] [INFO] [utils.py:789:see_memory_usage] CPU Virtual Memory: used = 77.67 GB, percent = 3.9% [2024-12-05 16:34:14,622] [INFO] [logging.py:129:log_dist] [Rank 0] DeepSpeed Final Optimizer = DeepSpeedZeroOptimizer [2024-12-05 16:34:14,622] [INFO] [logging.py:129:log_dist] [Rank 0] DeepSpeed using client callable to create LR scheduler [2024-12-05 16:34:14,623] [INFO] [logging.py:129:log_dist] [Rank 0] DeepSpeed LR Scheduler = PiecewiseLinearWarmupLR(warmup_steps_list=[0, 30000, 60000], warmup_lr_list=[0.0, 5e-05, 0.0002]) [2024-12-05 16:34:14,623] [INFO] [logging.py:129:log_dist] [Rank 0] step=0, skipped=0, lr=[np.float64(1.6666666666666667e-09)], mom=[[0.9, 0.98]] [2024-12-05 16:34:14,624] [INFO] [config.py:999:print] DeepSpeedEngine configuration: [2024-12-05 16:34:14,624] [INFO] [config.py:1003:print] activation_checkpointing_config { "partition_activations": false, "contiguous_memory_optimization": false, "cpu_checkpointing": false, "number_checkpoints": null, "synchronize_checkpoint_boundary": false, "profile": false } [2024-12-05 16:34:14,624] [INFO] [config.py:1003:print] aio_config ................... {'block_size': 1048576, 'queue_depth': 8, 'thread_count': 1, 'single_submit': False, 'overlap_events': True, 'use_gds': False} [2024-12-05 16:34:14,624] [INFO] [config.py:1003:print] amp_enabled .................. False [2024-12-05 16:34:14,624] [INFO] [config.py:1003:print] amp_params ................... False [2024-12-05 16:34:14,624] [INFO] [config.py:1003:print] autotuning_config ............ { "enabled": false, "start_step": null, "end_step": null, "metric_path": null, "arg_mappings": null, "metric": "throughput", "model_info": null, "results_dir": "autotuning_results", "exps_dir": "autotuning_exps", "overwrite": true, "fast": true, "start_profile_step": 3, "end_profile_step": 5, "tuner_type": "gridsearch", "tuner_early_stopping": 5, "tuner_num_trials": 50, "model_info_path": null, "mp_size": 1, "max_train_batch_size": null, "min_train_batch_size": 1, "max_train_micro_batch_size_per_gpu": 1.024000e+03, "min_train_micro_batch_size_per_gpu": 1, "num_tuning_micro_batch_sizes": 3 } [2024-12-05 16:34:14,624] [INFO] [config.py:1003:print] bfloat16_enabled ............. True [2024-12-05 16:34:14,624] [INFO] [config.py:1003:print] bfloat16_immediate_grad_update False [2024-12-05 16:34:14,624] [INFO] [config.py:1003:print] checkpoint_parallel_write_pipeline False [2024-12-05 16:34:14,624] [INFO] [config.py:1003:print] checkpoint_tag_validation_enabled True [2024-12-05 16:34:14,624] [INFO] [config.py:1003:print] checkpoint_tag_validation_fail False [2024-12-05 16:34:14,624] [INFO] [config.py:1003:print] comms_config ................. [2024-12-05 16:34:14,624] [INFO] [config.py:1003:print] communication_data_type ...... None [2024-12-05 16:34:14,624] [INFO] [config.py:1003:print] compression_config ........... {'weight_quantization': {'shared_parameters': {'enabled': False, 'quantizer_kernel': False, 'schedule_offset': 0, 'quantize_groups': 1, 'quantize_verbose': False, 'quantization_type': 'symmetric', 'quantize_weight_in_forward': False, 'rounding': 'nearest', 'fp16_mixed_quantize': False, 'quantize_change_ratio': 0.001}, 'different_groups': {}}, 'activation_quantization': {'shared_parameters': {'enabled': False, 'quantization_type': 'symmetric', 'range_calibration': 'dynamic', 'schedule_offset': 1000}, 'different_groups': {}}, 'sparse_pruning': {'shared_parameters': {'enabled': False, 'method': 'l1', 'schedule_offset': 1000}, 'different_groups': {}}, 'row_pruning': {'shared_parameters': {'enabled': False, 'method': 'l1', 'schedule_offset': 1000}, 'different_groups': {}}, 'head_pruning': {'shared_parameters': {'enabled': False, 'method': 'topk', 'schedule_offset': 1000}, 'different_groups': {}}, 'channel_pruning': {'shared_parameters': {'enabled': False, 'method': 'l1', 'schedule_offset': 1000}, 'different_groups': {}}, 'layer_reduction': {'enabled': False}} [2024-12-05 16:34:14,624] [INFO] [config.py:1003:print] curriculum_enabled_legacy .... False [2024-12-05 16:34:14,624] [INFO] [config.py:1003:print] curriculum_params_legacy ..... False [2024-12-05 16:34:14,624] [INFO] [config.py:1003:print] data_efficiency_config ....... {'enabled': False, 'seed': 1234, 'data_sampling': {'enabled': False, 'num_epochs': 1000, 'num_workers': 0, 'curriculum_learning': {'enabled': False}}, 'data_routing': {'enabled': False, 'random_ltd': {'enabled': False, 'layer_token_lr_schedule': {'enabled': False}}}} [2024-12-05 16:34:14,624] [INFO] [config.py:1003:print] data_efficiency_enabled ...... False [2024-12-05 16:34:14,624] [INFO] [config.py:1003:print] dataloader_drop_last ......... False [2024-12-05 16:34:14,624] [INFO] [config.py:1003:print] disable_allgather ............ False [2024-12-05 16:34:14,624] [INFO] [config.py:1003:print] dump_state ................... False [2024-12-05 16:34:14,624] [INFO] [config.py:1003:print] dynamic_loss_scale_args ...... None [2024-12-05 16:34:14,624] [INFO] [config.py:1003:print] eigenvalue_enabled ........... False [2024-12-05 16:34:14,624] [INFO] [config.py:1003:print] eigenvalue_gas_boundary_resolution 1 [2024-12-05 16:34:14,624] [INFO] [config.py:1003:print] eigenvalue_layer_name ........ bert.encoder.layer [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] eigenvalue_layer_num ......... 0 [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] eigenvalue_max_iter .......... 100 [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] eigenvalue_stability ......... 1e-06 [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] eigenvalue_tol ............... 0.01 [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] eigenvalue_verbose ........... False [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] elasticity_enabled ........... False [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] flops_profiler_config ........ { "enabled": false, "recompute_fwd_factor": 0.0, "profile_step": 1, "module_depth": -1, "top_modules": 1, "detailed": true, "output_file": null } [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] fp16_auto_cast ............... None [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] fp16_enabled ................. False [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] fp16_master_weights_and_gradients False [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] global_rank .................. 0 [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] grad_accum_dtype ............. None [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] gradient_accumulation_steps .. 1 [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] gradient_clipping ............ 5.0 [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] gradient_predivide_factor .... 1.0 [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] graph_harvesting ............. False [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] hybrid_engine ................ enabled=False max_out_tokens=512 inference_tp_size=1 release_inference_cache=False pin_parameters=True tp_gather_partition_size=8 [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] initial_dynamic_scale ........ 1 [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] load_universal_checkpoint .... False [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] loss_scale ................... 1.0 [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] memory_breakdown ............. False [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] mics_hierarchial_params_gather False [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] mics_shard_size .............. -1 [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] monitor_config ............... tensorboard=TensorBoardConfig(enabled=False, output_path='', job_name='DeepSpeedJobName') comet=CometConfig(enabled=False, samples_log_interval=100, project=None, workspace=None, api_key=None, experiment_name=None, experiment_key=None, online=None, mode=None) wandb=WandbConfig(enabled=False, group=None, team=None, project='deepspeed') csv_monitor=CSVConfig(enabled=False, output_path='', job_name='DeepSpeedJobName') [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] nebula_config ................ { "enabled": false, "persistent_storage_path": null, "persistent_time_interval": 100, "num_of_version_in_retention": 2, "enable_nebula_load": true, "load_path": null } [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] optimizer_legacy_fusion ...... False [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] optimizer_name ............... adam [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] optimizer_params ............. {'lr': 0.0002, 'betas': [0.9, 0.98], 'eps': 1e-06, 'weight_decay': 0.0} [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] pipeline ..................... {'stages': 'auto', 'partition': 'best', 'seed_layers': False, 'activation_checkpoint_interval': 0, 'pipe_partitioned': True, 'grad_partitioned': True} [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] pld_enabled .................. False [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] pld_params ................... False [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] prescale_gradients ........... False [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] scheduler_name ............... None [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] scheduler_params ............. None [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] seq_parallel_communication_data_type torch.float32 [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] sparse_attention ............. None [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] sparse_gradients_enabled ..... False [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] steps_per_print .............. 1000 [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] timers_config ................ enabled=True synchronized=True [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] train_batch_size ............. 16 [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] train_micro_batch_size_per_gpu 1 [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] use_data_before_expert_parallel_ False [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] use_node_local_storage ....... False [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] wall_clock_breakdown ......... False [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] weight_quantization_config ... None [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] world_size ................... 16 [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] zero_allow_untested_optimizer False [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] zero_config .................. stage=2 contiguous_gradients=True reduce_scatter=True reduce_bucket_size=500000000 use_multi_rank_bucket_allreduce=True allgather_partitions=True allgather_bucket_size=500000000 overlap_comm=True load_from_fp32_weights=True elastic_checkpoint=False offload_param=None offload_optimizer=None sub_group_size=1000000000 cpu_offload_param=None cpu_offload_use_pin_memory=None cpu_offload=None prefetch_bucket_size=50000000 param_persistence_threshold=100000 model_persistence_threshold=9223372036854775807 max_live_parameters=1000000000 max_reuse_distance=1000000000 gather_16bit_weights_on_model_save=False use_all_reduce_for_fetch_params=False stage3_gather_fp16_weights_on_model_save=False ignore_unused_parameters=True legacy_stage1=False round_robin_gradients=False zero_hpz_partition_size=1 zero_quantized_weights=False zero_quantized_nontrainable_weights=False zero_quantized_gradients=False mics_shard_size=-1 mics_hierarchical_params_gather=False memory_efficient_linear=True pipeline_loading_checkpoint=False override_module_apply=True [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] zero_enabled ................. True [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] zero_force_ds_cpu_optimizer .. True [2024-12-05 16:34:14,625] [INFO] [config.py:1003:print] zero_optimization_stage ...... 2 [2024-12-05 16:34:14,625] [INFO] [config.py:989:print_user_config] json = { "train_micro_batch_size_per_gpu": 1, "gradient_accumulation_steps": 1, "gradient_clipping": 5.0, "bf16": { "enabled": true }, "zero_optimization": { "stage": 2, "contiguous_gradients": true, "overlap_comm": true, "reduce_scatter": true, "reduce_bucket_size": 5.000000e+08, "allgather_bucket_size": 5.000000e+08 }, "optimizer": { "type": "Adam", "params": { "lr": 0.0002, "betas": [0.9, 0.98], "eps": 1e-06, "weight_decay": 0.0 } }, "wall_clock_breakdown": false, "steps_per_print": 1000 } [cnode7-012:0/16] 2024-12-05 16:34:14,627 (deepspeed_trainer:75) INFO: Resume training from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12 [2024-12-05 16:34:14,628] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt... /mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/site-packages/deepspeed/runtime/checkpoint_engine/torch_checkpoint_engine.py:28: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature. partition = torch.load(path, map_location=map_location) [2024-12-05 16:34:15,253] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt. [2024-12-05 16:34:15,253] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt. [2024-12-05 16:34:15,253] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt. [2024-12-05 16:34:15,253] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt. [2024-12-05 16:34:15,253] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt. [2024-12-05 16:34:15,253] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt. [2024-12-05 16:34:15,253] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt. [2024-12-05 16:34:15,253] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt. [2024-12-05 16:34:15,292] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt... [2024-12-05 16:34:15,293] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt... [2024-12-05 16:34:15,448] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt. [2024-12-05 16:34:15,448] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt. [2024-12-05 16:34:15,448] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt. [2024-12-05 16:34:15,448] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt. [2024-12-05 16:34:15,448] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt. [2024-12-05 16:34:15,448] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt. [2024-12-05 16:34:15,448] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt. [2024-12-05 16:34:15,448] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt. [2024-12-05 16:34:15,328] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt... [2024-12-05 16:34:15,329] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt... [2024-12-05 16:34:15,330] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt... [2024-12-05 16:34:15,331] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt... [2024-12-05 16:34:15,331] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt... [2024-12-05 16:34:15,331] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt... [2024-12-05 16:34:15,500] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt... [2024-12-05 16:34:15,537] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt... [2024-12-05 16:34:15,537] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt... [2024-12-05 16:34:15,538] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt... [2024-12-05 16:34:15,538] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt... [2024-12-05 16:34:15,539] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt... [2024-12-05 16:34:15,540] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt... [2024-12-05 16:34:15,541] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt... [2024-12-05 16:34:15,875] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt. [2024-12-05 16:34:15,875] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt. [2024-12-05 16:34:15,920] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt. [2024-12-05 16:34:15,920] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt. [2024-12-05 16:34:15,920] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt. [2024-12-05 16:34:15,920] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt. [2024-12-05 16:34:15,921] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt. [2024-12-05 16:34:15,922] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt. [2024-12-05 16:34:16,130] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt. [2024-12-05 16:34:16,146] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt. [2024-12-05 16:34:16,146] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt. [2024-12-05 16:34:16,146] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt. [2024-12-05 16:34:16,146] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt. [2024-12-05 16:34:16,147] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt. [2024-12-05 16:34:16,148] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt. [2024-12-05 16:34:16,149] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/mp_rank_00_model_states.pt. [2024-12-05 16:34:16,077] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt... [2024-12-05 16:34:16,078] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt... [2024-12-05 16:34:16,169] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt... [2024-12-05 16:34:16,169] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt... [2024-12-05 16:34:16,171] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt... [2024-12-05 16:34:16,174] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt... [2024-12-05 16:34:16,181] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt... [2024-12-05 16:34:16,204] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt... [2024-12-05 16:34:16,405] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/bf16_zero_pp_rank_8_mp_rank_00_optim_states.pt... [2024-12-05 16:34:16,414] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/bf16_zero_pp_rank_13_mp_rank_00_optim_states.pt... [2024-12-05 16:34:16,414] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/bf16_zero_pp_rank_12_mp_rank_00_optim_states.pt... [2024-12-05 16:34:16,419] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/bf16_zero_pp_rank_9_mp_rank_00_optim_states.pt... [2024-12-05 16:34:16,426] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/bf16_zero_pp_rank_14_mp_rank_00_optim_states.pt... [2024-12-05 16:34:16,428] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/bf16_zero_pp_rank_11_mp_rank_00_optim_states.pt... [2024-12-05 16:34:16,429] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/bf16_zero_pp_rank_10_mp_rank_00_optim_states.pt... [2024-12-05 16:34:16,452] [INFO] [torch_checkpoint_engine.py:27:load] [Torch] Loading checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/bf16_zero_pp_rank_15_mp_rank_00_optim_states.pt... [2024-12-05 16:34:16,504] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt. [2024-12-05 16:34:16,504] [INFO] [engine.py:3076:_get_all_zero_checkpoint_state_dicts] successfully read 16 ZeRO state_dicts for rank 5 [2024-12-05 16:34:16,523] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt. [2024-12-05 16:34:16,523] [INFO] [engine.py:3076:_get_all_zero_checkpoint_state_dicts] successfully read 16 ZeRO state_dicts for rank 2 [2024-12-05 16:34:16,565] [INFO] [engine.py:3026:_load_zero_checkpoint] loading 16 zero partition checkpoints for rank 5 [2024-12-05 16:34:16,582] [INFO] [engine.py:3026:_load_zero_checkpoint] loading 16 zero partition checkpoints for rank 2 [2024-12-05 16:34:16,583] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt. [2024-12-05 16:34:16,584] [INFO] [engine.py:3076:_get_all_zero_checkpoint_state_dicts] successfully read 16 ZeRO state_dicts for rank 4 [2024-12-05 16:34:16,601] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt. [2024-12-05 16:34:16,602] [INFO] [engine.py:3076:_get_all_zero_checkpoint_state_dicts] successfully read 16 ZeRO state_dicts for rank 0 [2024-12-05 16:34:16,643] [INFO] [engine.py:3026:_load_zero_checkpoint] loading 16 zero partition checkpoints for rank 4 [2024-12-05 16:34:16,675] [INFO] [engine.py:3026:_load_zero_checkpoint] loading 16 zero partition checkpoints for rank 0 [2024-12-05 16:34:16,681] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt. [2024-12-05 16:34:16,681] [INFO] [engine.py:3076:_get_all_zero_checkpoint_state_dicts] successfully read 16 ZeRO state_dicts for rank 3 [cnode7-012:0/16] 2024-12-05 16:34:16,684 (multiple_iter_factory:32) INFO: Building 0th iter-factory... [2024-12-05 16:34:16,721] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt. [2024-12-05 16:34:16,721] [INFO] [engine.py:3076:_get_all_zero_checkpoint_state_dicts] successfully read 16 ZeRO state_dicts for rank 7 [2024-12-05 16:34:16,878] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/bf16_zero_pp_rank_13_mp_rank_00_optim_states.pt. [2024-12-05 16:34:16,878] [INFO] [engine.py:3076:_get_all_zero_checkpoint_state_dicts] successfully read 16 ZeRO state_dicts for rank 13 [2024-12-05 16:34:16,884] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/bf16_zero_pp_rank_8_mp_rank_00_optim_states.pt. [2024-12-05 16:34:16,884] [INFO] [engine.py:3076:_get_all_zero_checkpoint_state_dicts] successfully read 16 ZeRO state_dicts for rank 8 [2024-12-05 16:34:16,740] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt. [2024-12-05 16:34:16,741] [INFO] [engine.py:3076:_get_all_zero_checkpoint_state_dicts] successfully read 16 ZeRO state_dicts for rank 6 [2024-12-05 16:34:16,742] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt. [2024-12-05 16:34:16,742] [INFO] [engine.py:3076:_get_all_zero_checkpoint_state_dicts] successfully read 16 ZeRO state_dicts for rank 1 [2024-12-05 16:34:16,753] [INFO] [engine.py:3026:_load_zero_checkpoint] loading 16 zero partition checkpoints for rank 3 [2024-12-05 16:34:16,778] [INFO] [engine.py:3026:_load_zero_checkpoint] loading 16 zero partition checkpoints for rank 7 [2024-12-05 16:34:16,928] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/bf16_zero_pp_rank_11_mp_rank_00_optim_states.pt. [2024-12-05 16:34:16,928] [INFO] [engine.py:3076:_get_all_zero_checkpoint_state_dicts] successfully read 16 ZeRO state_dicts for rank 11 [2024-12-05 16:34:16,805] [INFO] [engine.py:3026:_load_zero_checkpoint] loading 16 zero partition checkpoints for rank 6 [2024-12-05 16:34:16,951] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/bf16_zero_pp_rank_9_mp_rank_00_optim_states.pt. [2024-12-05 16:34:16,951] [INFO] [engine.py:3076:_get_all_zero_checkpoint_state_dicts] successfully read 16 ZeRO state_dicts for rank 9 [2024-12-05 16:34:16,954] [INFO] [engine.py:3026:_load_zero_checkpoint] loading 16 zero partition checkpoints for rank 13 [2024-12-05 16:34:16,961] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/bf16_zero_pp_rank_14_mp_rank_00_optim_states.pt. [2024-12-05 16:34:16,962] [INFO] [engine.py:3076:_get_all_zero_checkpoint_state_dicts] successfully read 16 ZeRO state_dicts for rank 14 [2024-12-05 16:34:16,968] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/bf16_zero_pp_rank_12_mp_rank_00_optim_states.pt. [2024-12-05 16:34:16,968] [INFO] [engine.py:3076:_get_all_zero_checkpoint_state_dicts] successfully read 16 ZeRO state_dicts for rank 12 [2024-12-05 16:34:16,831] [INFO] [engine.py:3026:_load_zero_checkpoint] loading 16 zero partition checkpoints for rank 1 [2024-12-05 16:34:16,978] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/bf16_zero_pp_rank_10_mp_rank_00_optim_states.pt. [2024-12-05 16:34:16,978] [INFO] [engine.py:3076:_get_all_zero_checkpoint_state_dicts] successfully read 16 ZeRO state_dicts for rank 10 [2024-12-05 16:34:17,005] [INFO] [engine.py:3026:_load_zero_checkpoint] loading 16 zero partition checkpoints for rank 11 [2024-12-05 16:34:17,011] [INFO] [engine.py:3026:_load_zero_checkpoint] loading 16 zero partition checkpoints for rank 8 [2024-12-05 16:34:17,027] [INFO] [engine.py:3026:_load_zero_checkpoint] loading 16 zero partition checkpoints for rank 9 [2024-12-05 16:34:17,036] [INFO] [engine.py:3026:_load_zero_checkpoint] loading 16 zero partition checkpoints for rank 14 [2024-12-05 16:34:17,036] [INFO] [engine.py:3026:_load_zero_checkpoint] loading 16 zero partition checkpoints for rank 12 [2024-12-05 16:34:17,043] [INFO] [torch_checkpoint_engine.py:29:load] [Torch] Loaded checkpoint from exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_12/12/bf16_zero_pp_rank_15_mp_rank_00_optim_states.pt. [2024-12-05 16:34:17,043] [INFO] [engine.py:3076:_get_all_zero_checkpoint_state_dicts] successfully read 16 ZeRO state_dicts for rank 15 [2024-12-05 16:34:17,076] [INFO] [engine.py:3026:_load_zero_checkpoint] loading 16 zero partition checkpoints for rank 10 [2024-12-05 16:34:17,184] [INFO] [engine.py:3026:_load_zero_checkpoint] loading 16 zero partition checkpoints for rank 15 [cnode7-012:0/16] 2024-12-05 16:34:42,799 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-05 16:34:59,567 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.4", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.4", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.4", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.4", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-05 16:34:59,567 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.4, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-05 16:34:59,570 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-05 16:35:20,760] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:35:20,814] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:35:21,029] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:35:23,959] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:35:24,432] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:35:24,769] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:35:25,001] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:35:25,065] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:35:25,311] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:35:25,406] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:35:25,825] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:35:25,924] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:35:26,094] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:35:26,214] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:35:26,158] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:35:37,871] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:36:05,909] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:36:06,562] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:36:09,120] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:36:12,029] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:36:12,590] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:36:12,769] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:36:13,377] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:36:13,470] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:36:13,475] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:36:13,747] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:36:14,028] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:36:14,088] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:36:14,606] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:36:14,873] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:36:14,918] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:36:27,668] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:36:50,773] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:36:51,679] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:36:56,056] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:37:00,541] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:37:00,932] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:37:01,433] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:37:01,582] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:37:02,107] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:37:02,028] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:37:02,077] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:37:02,436] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:37:02,367] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:37:02,897] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:37:03,503] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:37:03,592] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:37:14,499] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:37:34,999] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:37:35,943] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:37:42,928] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:37:47,677] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:37:48,717] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:37:48,993] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:37:49,909] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:37:50,158] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:37:50,381] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:37:50,420] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:37:50,425] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:37:51,504] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:37:52,944] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:37:54,547] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) /mnt/home/williamchen/espnet/espnet2/s2t/espnet_model.py:279: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. with autocast(False): /mnt/home/williamchen/espnet/espnet2/s2t/espnet_model.py:279: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. with autocast(False): [2024-12-05 16:38:01,538] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:38:01,725] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) /mnt/home/williamchen/espnet/espnet2/s2t/espnet_model.py:279: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. with autocast(False): /mnt/home/williamchen/espnet/espnet2/s2t/espnet_model.py:279: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. with autocast(False): cnode7-013:736481:744977 [0] NCCL INFO Using non-device net plugin version 0 cnode7-013:736481:744977 [0] NCCL INFO Using network IB cnode7-012:1272010:1281060 [5] NCCL INFO Using non-device net plugin version 0 cnode7-012:1272010:1281060 [5] NCCL INFO Using network IB cnode7-012:1272005:1281059 [0] NCCL INFO Using non-device net plugin version 0 cnode7-012:1272005:1281059 [0] NCCL INFO Using network IB cnode7-012:1272007:1281061 [2] NCCL INFO Using non-device net plugin version 0 cnode7-012:1272007:1281061 [2] NCCL INFO Using network IB /mnt/home/williamchen/espnet/espnet2/s2t/espnet_model.py:279: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. with autocast(False): /mnt/home/williamchen/espnet/espnet2/s2t/espnet_model.py:279: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. with autocast(False): cnode7-012:1272009:1281281 [4] NCCL INFO Using non-device net plugin version 0 cnode7-012:1272009:1281281 [4] NCCL INFO Using network IB /mnt/home/williamchen/espnet/espnet2/s2t/espnet_model.py:279: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. with autocast(False): cnode7-012:1272012:1281554 [7] NCCL INFO Using non-device net plugin version 0 cnode7-012:1272012:1281554 [7] NCCL INFO Using network IB /mnt/home/williamchen/espnet/espnet2/s2t/espnet_model.py:279: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. with autocast(False): /mnt/home/williamchen/espnet/espnet2/s2t/espnet_model.py:279: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. with autocast(False): cnode7-012:1272006:1281620 [1] NCCL INFO Using non-device net plugin version 0 cnode7-012:1272006:1281620 [1] NCCL INFO Using network IB /mnt/home/williamchen/espnet/espnet2/s2t/espnet_model.py:279: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. with autocast(False): /mnt/home/williamchen/espnet/espnet2/s2t/espnet_model.py:279: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. with autocast(False): cnode7-012:1272008:1281624 [3] NCCL INFO Using non-device net plugin version 0 cnode7-012:1272008:1281624 [3] NCCL INFO Using network IB cnode7-012:1272011:1281645 [6] NCCL INFO Using non-device net plugin version 0 cnode7-012:1272011:1281645 [6] NCCL INFO Using network IB cnode7-013:736485:745370 [4] NCCL INFO Using non-device net plugin version 0 cnode7-013:736485:745370 [4] NCCL INFO Using network IB cnode7-013:736484:745381 [3] NCCL INFO Using non-device net plugin version 0 cnode7-013:736484:745381 [3] NCCL INFO Using network IB /mnt/home/williamchen/espnet/espnet2/s2t/espnet_model.py:279: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. with autocast(False): cnode7-013:736483:745475 [2] NCCL INFO Using non-device net plugin version 0 cnode7-013:736483:745475 [2] NCCL INFO Using network IB /mnt/home/williamchen/espnet/espnet2/s2t/espnet_model.py:279: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. with autocast(False): cnode7-013:736486:745644 [5] NCCL INFO Using non-device net plugin version 0 cnode7-013:736486:745644 [5] NCCL INFO Using network IB /mnt/home/williamchen/espnet/espnet2/s2t/espnet_model.py:279: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. with autocast(False): cnode7-013:736488:746041 [7] NCCL INFO Using non-device net plugin version 0 cnode7-013:736488:746041 [7] NCCL INFO Using network IB /mnt/home/williamchen/espnet/espnet2/s2t/espnet_model.py:279: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. with autocast(False): cnode7-013:736487:746330 [6] NCCL INFO Using non-device net plugin version 0 cnode7-013:736487:746330 [6] NCCL INFO Using network IB /mnt/home/williamchen/espnet/espnet2/s2t/espnet_model.py:279: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. with autocast(False): cnode7-013:736488:746041 [7] NCCL INFO comm 0x5555cf9aad40 rank 15 nranks 16 cudaDev 7 nvmlDev 7 busId df000 commId 0x17e756722888e895 - Init START cnode7-012:1272008:1281624 [3] NCCL INFO comm 0x5555e39b2180 rank 3 nranks 16 cudaDev 3 nvmlDev 3 busId 61000 commId 0x17e756722888e895 - Init START cnode7-013:736482:746397 [1] NCCL INFO Using non-device net plugin version 0 cnode7-013:736482:746397 [1] NCCL INFO Using network IB cnode7-013:736482:746397 [1] NCCL INFO comm 0x5555cf9ca2c0 rank 9 nranks 16 cudaDev 1 nvmlDev 1 busId 43000 commId 0x17e756722888e895 - Init START cnode7-012:1272007:1281061 [2] NCCL INFO comm 0x5555e39b9f60 rank 2 nranks 16 cudaDev 2 nvmlDev 2 busId 52000 commId 0x17e756722888e895 - Init START cnode7-013:736484:745381 [3] NCCL INFO comm 0x5555cd20c9e0 rank 11 nranks 16 cudaDev 3 nvmlDev 3 busId 61000 commId 0x17e756722888e895 - Init START cnode7-013:736487:746330 [6] NCCL INFO comm 0x5555cf9ae150 rank 14 nranks 16 cudaDev 6 nvmlDev 6 busId d1000 commId 0x17e756722888e895 - Init START cnode7-013:736485:745370 [4] NCCL INFO comm 0x5555cf9ca4a0 rank 12 nranks 16 cudaDev 4 nvmlDev 4 busId 9d000 commId 0x17e756722888e895 - Init START cnode7-013:736481:744977 [0] NCCL INFO comm 0x5555c59f0f70 rank 8 nranks 16 cudaDev 0 nvmlDev 0 busId 1b000 commId 0x17e756722888e895 - Init START cnode7-013:736486:745644 [5] NCCL INFO comm 0x5555cd22bfa0 rank 13 nranks 16 cudaDev 5 nvmlDev 5 busId c3000 commId 0x17e756722888e895 - Init START cnode7-012:1272012:1281554 [7] NCCL INFO comm 0x5555cf991e20 rank 7 nranks 16 cudaDev 7 nvmlDev 7 busId df000 commId 0x17e756722888e895 - Init START cnode7-012:1272009:1281281 [4] NCCL INFO comm 0x5555cd223a50 rank 4 nranks 16 cudaDev 4 nvmlDev 4 busId 9d000 commId 0x17e756722888e895 - Init START cnode7-012:1272006:1281620 [1] NCCL INFO comm 0x5555cf9b5e30 rank 1 nranks 16 cudaDev 1 nvmlDev 1 busId 43000 commId 0x17e756722888e895 - Init START cnode7-012:1272011:1281645 [6] NCCL INFO comm 0x5555cf994130 rank 6 nranks 16 cudaDev 6 nvmlDev 6 busId d1000 commId 0x17e756722888e895 - Init START cnode7-013:736483:745475 [2] NCCL INFO comm 0x5555cf9ca470 rank 10 nranks 16 cudaDev 2 nvmlDev 2 busId 52000 commId 0x17e756722888e895 - Init START cnode7-012:1272010:1281060 [5] NCCL INFO comm 0x5555e39cd100 rank 5 nranks 16 cudaDev 5 nvmlDev 5 busId c3000 commId 0x17e756722888e895 - Init START cnode7-012:1272005:1281059 [0] NCCL INFO comm 0x5555c2f5fad0 rank 0 nranks 16 cudaDev 0 nvmlDev 0 busId 1b000 commId 0x17e756722888e895 - Init START cnode7-013:736486:745644 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffffff00,00000000,0000ffff,ffffffff,ff000000,00000000 cnode7-013:736486:745644 [5] NCCL INFO NVLS multicast support is available on dev 5 cnode7-013:736485:745370 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffffff00,00000000,0000ffff,ffffffff,ff000000,00000000 cnode7-013:736485:745370 [4] NCCL INFO NVLS multicast support is available on dev 4 cnode7-013:736483:745475 [2] NCCL INFO Setting affinity for GPU 2 to ff,ffffffff,ffff0000,00000000,00ffffff,ffffffff cnode7-013:736483:745475 [2] NCCL INFO NVLS multicast support is available on dev 2 cnode7-013:736488:746041 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffffff00,00000000,0000ffff,ffffffff,ff000000,00000000 cnode7-013:736488:746041 [7] NCCL INFO NVLS multicast support is available on dev 7 cnode7-013:736484:745381 [3] NCCL INFO Setting affinity for GPU 3 to ff,ffffffff,ffff0000,00000000,00ffffff,ffffffff cnode7-013:736484:745381 [3] NCCL INFO NVLS multicast support is available on dev 3 cnode7-012:1272012:1281554 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffffff00,00000000,0000ffff,ffffffff,ff000000,00000000 cnode7-012:1272012:1281554 [7] NCCL INFO NVLS multicast support is available on dev 7 cnode7-013:736482:746397 [1] NCCL INFO Setting affinity for GPU 1 to ff,ffffffff,ffff0000,00000000,00ffffff,ffffffff cnode7-013:736482:746397 [1] NCCL INFO NVLS multicast support is available on dev 1 cnode7-013:736487:746330 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffffff00,00000000,0000ffff,ffffffff,ff000000,00000000 cnode7-013:736487:746330 [6] NCCL INFO NVLS multicast support is available on dev 6 cnode7-013:736481:744977 [0] NCCL INFO Setting affinity for GPU 0 to ff,ffffffff,ffff0000,00000000,00ffffff,ffffffff cnode7-013:736481:744977 [0] NCCL INFO NVLS multicast support is available on dev 0 cnode7-012:1272009:1281281 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffffff00,00000000,0000ffff,ffffffff,ff000000,00000000 cnode7-012:1272009:1281281 [4] NCCL INFO NVLS multicast support is available on dev 4 cnode7-012:1272006:1281620 [1] NCCL INFO Setting affinity for GPU 1 to ff,ffffffff,ffff0000,00000000,00ffffff,ffffffff cnode7-012:1272006:1281620 [1] NCCL INFO NVLS multicast support is available on dev 1 cnode7-012:1272010:1281060 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffffff00,00000000,0000ffff,ffffffff,ff000000,00000000 cnode7-012:1272010:1281060 [5] NCCL INFO NVLS multicast support is available on dev 5 cnode7-012:1272005:1281059 [0] NCCL INFO Setting affinity for GPU 0 to ff,ffffffff,ffff0000,00000000,00ffffff,ffffffff cnode7-012:1272005:1281059 [0] NCCL INFO NVLS multicast support is available on dev 0 cnode7-012:1272008:1281624 [3] NCCL INFO Setting affinity for GPU 3 to ff,ffffffff,ffff0000,00000000,00ffffff,ffffffff cnode7-012:1272008:1281624 [3] NCCL INFO NVLS multicast support is available on dev 3 cnode7-012:1272007:1281061 [2] NCCL INFO Setting affinity for GPU 2 to ff,ffffffff,ffff0000,00000000,00ffffff,ffffffff cnode7-012:1272007:1281061 [2] NCCL INFO NVLS multicast support is available on dev 2 cnode7-013:736486:745644 [5] NCCL INFO comm 0x5555cd22bfa0 rank 13 nRanks 16 nNodes 2 localRanks 8 localRank 5 MNNVL 0 cnode7-013:736486:745644 [5] NCCL INFO Trees [0] 14/-1/-1->13->12 [1] 14/-1/-1->13->12 [2] 14/-1/-1->13->12 [3] 14/-1/-1->13->12 [4] 14/-1/-1->13->12 [5] 14/-1/-1->13->5 [6] -1/-1/-1->13->12 [7] 14/-1/-1->13->12 [8] 14/-1/-1->13->12 [9] 14/-1/-1->13->12 [10] 14/-1/-1->13->12 [11] 14/-1/-1->13->12 [12] 14/-1/-1->13->12 [13] 14/5/-1->13->-1 [14] -1/-1/-1->13->12 [15] 14/-1/-1->13->12 cnode7-013:736486:745644 [5] NCCL INFO P2P Chunksize set to 131072 cnode7-013:736488:746041 [7] NCCL INFO comm 0x5555cf9aad40 rank 15 nRanks 16 nNodes 2 localRanks 8 localRank 7 MNNVL 0 cnode7-013:736488:746041 [7] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] 8/-1/-1->15->14 [2] 8/-1/-1->15->14 [3] 8/-1/-1->15->14 [4] 8/-1/-1->15->14 [5] 8/-1/-1->15->14 [6] 8/-1/-1->15->14 [7] 8/-1/-1->15->7 [8] -1/-1/-1->15->14 [9] 8/-1/-1->15->14 [10] 8/-1/-1->15->14 [11] 8/-1/-1->15->14 [12] 8/-1/-1->15->14 [13] 8/-1/-1->15->14 [14] 8/-1/-1->15->14 [15] 8/7/-1->15->-1 cnode7-013:736488:746041 [7] NCCL INFO P2P Chunksize set to 131072 cnode7-013:736485:745370 [4] NCCL INFO comm 0x5555cf9ca4a0 rank 12 nRanks 16 nNodes 2 localRanks 8 localRank 4 MNNVL 0 cnode7-013:736485:745370 [4] NCCL INFO Trees [0] 13/-1/-1->12->11 [1] 13/-1/-1->12->11 [2] 13/-1/-1->12->11 [3] 13/-1/-1->12->11 [4] 13/-1/-1->12->4 [5] -1/-1/-1->12->11 [6] 13/-1/-1->12->11 [7] 13/-1/-1->12->11 [8] 13/-1/-1->12->11 [9] 13/-1/-1->12->11 [10] 13/-1/-1->12->11 [11] 13/-1/-1->12->11 [12] 13/4/-1->12->-1 [13] -1/-1/-1->12->11 [14] 13/-1/-1->12->11 [15] 13/-1/-1->12->11 cnode7-013:736485:745370 [4] NCCL INFO P2P Chunksize set to 131072 cnode7-013:736487:746330 [6] NCCL INFO comm 0x5555cf9ae150 rank 14 nRanks 16 nNodes 2 localRanks 8 localRank 6 MNNVL 0 cnode7-013:736487:746330 [6] NCCL INFO Trees [0] 15/-1/-1->14->13 [1] 15/-1/-1->14->13 [2] 15/-1/-1->14->13 [3] 15/-1/-1->14->13 [4] 15/-1/-1->14->13 [5] 15/-1/-1->14->13 [6] 15/-1/-1->14->6 [7] -1/-1/-1->14->13 [8] 15/-1/-1->14->13 [9] 15/-1/-1->14->13 [10] 15/-1/-1->14->13 [11] 15/-1/-1->14->13 [12] 15/-1/-1->14->13 [13] 15/-1/-1->14->13 [14] 15/6/-1->14->-1 [15] -1/-1/-1->14->13 cnode7-013:736487:746330 [6] NCCL INFO P2P Chunksize set to 131072 cnode7-012:1272006:1281620 [1] NCCL INFO comm 0x5555cf9b5e30 rank 1 nRanks 16 nNodes 2 localRanks 8 localRank 1 MNNVL 0 cnode7-012:1272006:1281620 [1] NCCL INFO NVLS Head 0: 0 8 cnode7-012:1272006:1281620 [1] NCCL INFO NVLS Head 1: 1 9 cnode7-012:1272006:1281620 [1] NCCL INFO NVLS Head 2: 2 10 cnode7-012:1272006:1281620 [1] NCCL INFO NVLS Head 3: 3 11 cnode7-012:1272006:1281620 [1] NCCL INFO NVLS Head 4: 4 12 cnode7-012:1272006:1281620 [1] NCCL INFO NVLS Head 5: 5 13 cnode7-012:1272006:1281620 [1] NCCL INFO NVLS Head 6: 6 14 cnode7-012:1272006:1281620 [1] NCCL INFO NVLS Head 7: 7 15 cnode7-012:1272006:1281620 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/9/-1->1->-1 [2] -1/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->9 [10] -1/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 cnode7-012:1272006:1281620 [1] NCCL INFO P2P Chunksize set to 131072 cnode7-012:1272007:1281061 [2] NCCL INFO comm 0x5555e39b9f60 rank 2 nRanks 16 nNodes 2 localRanks 8 localRank 2 MNNVL 0 cnode7-012:1272007:1281061 [2] NCCL INFO NVLS Head 0: 0 8 cnode7-012:1272007:1281061 [2] NCCL INFO NVLS Head 1: 1 9 cnode7-012:1272007:1281061 [2] NCCL INFO NVLS Head 2: 2 10 cnode7-012:1272007:1281061 [2] NCCL INFO NVLS Head 3: 3 11 cnode7-012:1272007:1281061 [2] NCCL INFO NVLS Head 4: 4 12 cnode7-012:1272007:1281061 [2] NCCL INFO NVLS Head 5: 5 13 cnode7-012:1272007:1281061 [2] NCCL INFO NVLS Head 6: 6 14 cnode7-012:1272007:1281061 [2] NCCL INFO NVLS Head 7: 7 15 cnode7-012:1272007:1281061 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/10/-1->2->-1 [3] -1/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->10 [11] -1/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 cnode7-012:1272007:1281061 [2] NCCL INFO P2P Chunksize set to 131072 cnode7-012:1272012:1281554 [7] NCCL INFO comm 0x5555cf991e20 rank 7 nRanks 16 nNodes 2 localRanks 8 localRank 7 MNNVL 0 cnode7-012:1272012:1281554 [7] NCCL INFO NVLS Head 0: 0 8 cnode7-012:1272012:1281554 [7] NCCL INFO NVLS Head 1: 1 9 cnode7-012:1272012:1281554 [7] NCCL INFO NVLS Head 2: 2 10 cnode7-012:1272012:1281554 [7] NCCL INFO NVLS Head 3: 3 11 cnode7-012:1272012:1281554 [7] NCCL INFO NVLS Head 4: 4 12 cnode7-012:1272012:1281554 [7] NCCL INFO NVLS Head 5: 5 13 cnode7-012:1272012:1281554 [7] NCCL INFO NVLS Head 6: 6 14 cnode7-012:1272012:1281554 [7] NCCL INFO NVLS Head 7: 7 15 cnode7-012:1272012:1281554 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] 0/-1/-1->7->6 [2] 0/-1/-1->7->6 [3] 0/-1/-1->7->6 [4] 0/-1/-1->7->6 [5] 0/-1/-1->7->6 [6] 0/-1/-1->7->6 [7] 0/15/-1->7->-1 [8] -1/-1/-1->7->6 [9] 0/-1/-1->7->6 [10] 0/-1/-1->7->6 [11] 0/-1/-1->7->6 [12] 0/-1/-1->7->6 [13] 0/-1/-1->7->6 [14] 0/-1/-1->7->6 [15] 0/-1/-1->7->15 cnode7-012:1272012:1281554 [7] NCCL INFO P2P Chunksize set to 131072 cnode7-012:1272005:1281059 [0] NCCL INFO comm 0x5555c2f5fad0 rank 0 nRanks 16 nNodes 2 localRanks 8 localRank 0 MNNVL 0 cnode7-012:1272005:1281059 [0] NCCL INFO NVLS Head 0: 0 8 cnode7-012:1272005:1281059 [0] NCCL INFO NVLS Head 1: 1 9 cnode7-012:1272005:1281059 [0] NCCL INFO NVLS Head 2: 2 10 cnode7-012:1272005:1281059 [0] NCCL INFO NVLS Head 3: 3 11 cnode7-012:1272005:1281059 [0] NCCL INFO NVLS Head 4: 4 12 cnode7-012:1272005:1281059 [0] NCCL INFO NVLS Head 5: 5 13 cnode7-012:1272005:1281059 [0] NCCL INFO NVLS Head 6: 6 14 cnode7-012:1272005:1281059 [0] NCCL INFO NVLS Head 7: 7 15 cnode7-012:1272005:1281059 [0] NCCL INFO Channel 00/16 : 0 7 6 5 4 3 2 1 9 10 11 12 13 14 15 8 cnode7-012:1272005:1281059 [0] NCCL INFO Channel 01/16 : 0 8 15 14 13 12 11 10 9 1 2 3 4 5 6 7 cnode7-012:1272005:1281059 [0] NCCL INFO Channel 02/16 : 0 7 6 5 4 3 11 12 13 14 15 8 9 10 2 1 cnode7-012:1272005:1281059 [0] NCCL INFO Channel 03/16 : 0 1 2 10 9 8 15 14 13 12 11 3 4 5 6 7 cnode7-012:1272005:1281059 [0] NCCL INFO Channel 04/16 : 0 7 6 5 13 14 15 8 9 10 11 12 4 3 2 1 cnode7-012:1272005:1281059 [0] NCCL INFO Channel 05/16 : 0 1 2 3 4 12 11 10 9 8 15 14 13 5 6 7 cnode7-012:1272005:1281059 [0] NCCL INFO Channel 06/16 : 0 7 15 8 9 10 11 12 13 14 6 5 4 3 2 1 cnode7-012:1272005:1281059 [0] NCCL INFO Channel 07/16 : 0 1 2 3 4 5 6 14 13 12 11 10 9 8 15 7 cnode7-012:1272005:1281059 [0] NCCL INFO Channel 08/16 : 0 7 6 5 4 3 2 1 9 10 11 12 13 14 15 8 cnode7-012:1272005:1281059 [0] NCCL INFO Channel 09/16 : 0 8 15 14 13 12 11 10 9 1 2 3 4 5 6 7 cnode7-012:1272005:1281059 [0] NCCL INFO Channel 10/16 : 0 7 6 5 4 3 11 12 13 14 15 8 9 10 2 1 cnode7-012:1272005:1281059 [0] NCCL INFO Channel 11/16 : 0 1 2 10 9 8 15 14 13 12 11 3 4 5 6 7 cnode7-012:1272005:1281059 [0] NCCL INFO Channel 12/16 : 0 7 6 5 13 14 15 8 9 10 11 12 4 3 2 1 cnode7-012:1272005:1281059 [0] NCCL INFO Channel 13/16 : 0 1 2 3 4 12 11 10 9 8 15 14 13 5 6 7 cnode7-012:1272005:1281059 [0] NCCL INFO Channel 14/16 : 0 7 15 8 9 10 11 12 13 14 6 5 4 3 2 1 cnode7-012:1272005:1281059 [0] NCCL INFO Channel 15/16 : 0 1 2 3 4 5 6 14 13 12 11 10 9 8 15 7 cnode7-012:1272005:1281059 [0] NCCL INFO Trees [0] 1/8/-1->0->-1 [1] -1/-1/-1->0->7 [2] 1/-1/-1->0->7 [3] 1/-1/-1->0->7 [4] 1/-1/-1->0->7 [5] 1/-1/-1->0->7 [6] 1/-1/-1->0->7 [7] 1/-1/-1->0->7 [8] 1/-1/-1->0->8 [9] -1/-1/-1->0->7 [10] 1/-1/-1->0->7 [11] 1/-1/-1->0->7 [12] 1/-1/-1->0->7 [13] 1/-1/-1->0->7 [14] 1/-1/-1->0->7 [15] 1/-1/-1->0->7 cnode7-012:1272005:1281059 [0] NCCL INFO P2P Chunksize set to 131072 cnode7-012:1272011:1281645 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffffff00,00000000,0000ffff,ffffffff,ff000000,00000000 cnode7-012:1272011:1281645 [6] NCCL INFO NVLS multicast support is available on dev 6 cnode7-012:1272011:1281645 [6] NCCL INFO comm 0x5555cf994130 rank 6 nRanks 16 nNodes 2 localRanks 8 localRank 6 MNNVL 0 cnode7-012:1272011:1281645 [6] NCCL INFO NVLS Head 0: 0 8 cnode7-012:1272011:1281645 [6] NCCL INFO NVLS Head 1: 1 9 cnode7-012:1272011:1281645 [6] NCCL INFO NVLS Head 2: 2 10 cnode7-012:1272011:1281645 [6] NCCL INFO NVLS Head 3: 3 11 cnode7-012:1272011:1281645 [6] NCCL INFO NVLS Head 4: 4 12 cnode7-012:1272011:1281645 [6] NCCL INFO NVLS Head 5: 5 13 cnode7-012:1272011:1281645 [6] NCCL INFO NVLS Head 6: 6 14 cnode7-012:1272011:1281645 [6] NCCL INFO NVLS Head 7: 7 15 cnode7-012:1272011:1281645 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/14/-1->6->-1 [7] -1/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->14 [15] -1/-1/-1->6->5 cnode7-012:1272011:1281645 [6] NCCL INFO P2P Chunksize set to 131072 cnode7-012:1272010:1281060 [5] NCCL INFO comm 0x5555e39cd100 rank 5 nRanks 16 nNodes 2 localRanks 8 localRank 5 MNNVL 0 cnode7-012:1272010:1281060 [5] NCCL INFO NVLS Head 0: 0 8 cnode7-012:1272010:1281060 [5] NCCL INFO NVLS Head 1: 1 9 cnode7-012:1272010:1281060 [5] NCCL INFO NVLS Head 2: 2 10 cnode7-012:1272010:1281060 [5] NCCL INFO NVLS Head 3: 3 11 cnode7-012:1272010:1281060 [5] NCCL INFO NVLS Head 4: 4 12 cnode7-012:1272010:1281060 [5] NCCL INFO NVLS Head 5: 5 13 cnode7-012:1272010:1281060 [5] NCCL INFO NVLS Head 6: 6 14 cnode7-012:1272010:1281060 [5] NCCL INFO NVLS Head 7: 7 15 cnode7-012:1272010:1281060 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/13/-1->5->-1 [6] -1/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->13 [14] -1/-1/-1->5->4 [15] 6/-1/-1->5->4 cnode7-012:1272010:1281060 [5] NCCL INFO P2P Chunksize set to 131072 cnode7-012:1272009:1281281 [4] NCCL INFO comm 0x5555cd223a50 rank 4 nRanks 16 nNodes 2 localRanks 8 localRank 4 MNNVL 0 cnode7-012:1272009:1281281 [4] NCCL INFO NVLS Head 0: 0 8 cnode7-012:1272009:1281281 [4] NCCL INFO NVLS Head 1: 1 9 cnode7-012:1272009:1281281 [4] NCCL INFO NVLS Head 2: 2 10 cnode7-012:1272009:1281281 [4] NCCL INFO NVLS Head 3: 3 11 cnode7-012:1272009:1281281 [4] NCCL INFO NVLS Head 4: 4 12 cnode7-012:1272009:1281281 [4] NCCL INFO NVLS Head 5: 5 13 cnode7-012:1272009:1281281 [4] NCCL INFO NVLS Head 6: 6 14 cnode7-012:1272009:1281281 [4] NCCL INFO NVLS Head 7: 7 15 cnode7-012:1272009:1281281 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/12/-1->4->-1 [5] -1/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->12 [13] -1/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 cnode7-012:1272009:1281281 [4] NCCL INFO P2P Chunksize set to 131072 cnode7-012:1272008:1281624 [3] NCCL INFO comm 0x5555e39b2180 rank 3 nRanks 16 nNodes 2 localRanks 8 localRank 3 MNNVL 0 cnode7-012:1272008:1281624 [3] NCCL INFO NVLS Head 0: 0 8 cnode7-012:1272008:1281624 [3] NCCL INFO NVLS Head 1: 1 9 cnode7-012:1272008:1281624 [3] NCCL INFO NVLS Head 2: 2 10 cnode7-012:1272008:1281624 [3] NCCL INFO NVLS Head 3: 3 11 cnode7-012:1272008:1281624 [3] NCCL INFO NVLS Head 4: 4 12 cnode7-012:1272008:1281624 [3] NCCL INFO NVLS Head 5: 5 13 cnode7-012:1272008:1281624 [3] NCCL INFO NVLS Head 6: 6 14 cnode7-012:1272008:1281624 [3] NCCL INFO NVLS Head 7: 7 15 cnode7-012:1272008:1281624 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/11/-1->3->-1 [4] -1/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->11 [12] -1/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 cnode7-012:1272008:1281624 [3] NCCL INFO P2P Chunksize set to 131072 cnode7-012:1272011:1281645 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/CUMEM cnode7-012:1272011:1281645 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/CUMEM cnode7-012:1272011:1281645 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/CUMEM cnode7-012:1272011:1281645 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/CUMEM cnode7-012:1272011:1281645 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/CUMEM cnode7-012:1272011:1281645 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/CUMEM cnode7-013:736484:745381 [3] NCCL INFO comm 0x5555cd20c9e0 rank 11 nRanks 16 nNodes 2 localRanks 8 localRank 3 MNNVL 0 cnode7-013:736484:745381 [3] NCCL INFO Trees [0] 12/-1/-1->11->10 [1] 12/-1/-1->11->10 [2] 12/-1/-1->11->10 [3] 12/-1/-1->11->3 [4] -1/-1/-1->11->10 [5] 12/-1/-1->11->10 [6] 12/-1/-1->11->10 [7] 12/-1/-1->11->10 [8] 12/-1/-1->11->10 [9] 12/-1/-1->11->10 [10] 12/-1/-1->11->10 [11] 12/3/-1->11->-1 [12] -1/-1/-1->11->10 [13] 12/-1/-1->11->10 [14] 12/-1/-1->11->10 [15] 12/-1/-1->11->10 cnode7-013:736484:745381 [3] NCCL INFO P2P Chunksize set to 131072 cnode7-013:736483:745475 [2] NCCL INFO comm 0x5555cf9ca470 rank 10 nRanks 16 nNodes 2 localRanks 8 localRank 2 MNNVL 0 cnode7-013:736483:745475 [2] NCCL INFO Trees [0] 11/-1/-1->10->9 [1] 11/-1/-1->10->9 [2] 11/-1/-1->10->2 [3] -1/-1/-1->10->9 [4] 11/-1/-1->10->9 [5] 11/-1/-1->10->9 [6] 11/-1/-1->10->9 [7] 11/-1/-1->10->9 [8] 11/-1/-1->10->9 [9] 11/-1/-1->10->9 [10] 11/2/-1->10->-1 [11] -1/-1/-1->10->9 [12] 11/-1/-1->10->9 [13] 11/-1/-1->10->9 [14] 11/-1/-1->10->9 [15] 11/-1/-1->10->9 cnode7-013:736483:745475 [2] NCCL INFO P2P Chunksize set to 131072 cnode7-013:736481:744977 [0] NCCL INFO comm 0x5555c59f0f70 rank 8 nRanks 16 nNodes 2 localRanks 8 localRank 0 MNNVL 0 cnode7-013:736481:744977 [0] NCCL INFO Trees [0] 9/-1/-1->8->0 [1] -1/-1/-1->8->15 [2] 9/-1/-1->8->15 [3] 9/-1/-1->8->15 [4] 9/-1/-1->8->15 [5] 9/-1/-1->8->15 [6] 9/-1/-1->8->15 [7] 9/-1/-1->8->15 [8] 9/0/-1->8->-1 [9] -1/-1/-1->8->15 [10] 9/-1/-1->8->15 [11] 9/-1/-1->8->15 [12] 9/-1/-1->8->15 [13] 9/-1/-1->8->15 [14] 9/-1/-1->8->15 [15] 9/-1/-1->8->15 cnode7-013:736481:744977 [0] NCCL INFO P2P Chunksize set to 131072 cnode7-013:736482:746397 [1] NCCL INFO comm 0x5555cf9ca2c0 rank 9 nRanks 16 nNodes 2 localRanks 8 localRank 1 MNNVL 0 cnode7-013:736482:746397 [1] NCCL INFO Trees [0] 10/-1/-1->9->8 [1] 10/-1/-1->9->1 [2] -1/-1/-1->9->8 [3] 10/-1/-1->9->8 [4] 10/-1/-1->9->8 [5] 10/-1/-1->9->8 [6] 10/-1/-1->9->8 [7] 10/-1/-1->9->8 [8] 10/-1/-1->9->8 [9] 10/1/-1->9->-1 [10] -1/-1/-1->9->8 [11] 10/-1/-1->9->8 [12] 10/-1/-1->9->8 [13] 10/-1/-1->9->8 [14] 10/-1/-1->9->8 [15] 10/-1/-1->9->8 cnode7-013:736482:746397 [1] NCCL INFO P2P Chunksize set to 131072 cnode7-013:736483:745475 [2] NCCL INFO Channel 00/0 : 10[2] -> 11[3] via P2P/CUMEM cnode7-013:736483:745475 [2] NCCL INFO Channel 04/0 : 10[2] -> 11[3] via P2P/CUMEM cnode7-013:736483:745475 [2] NCCL INFO Channel 06/0 : 10[2] -> 11[3] via P2P/CUMEM cnode7-013:736483:745475 [2] NCCL INFO Channel 08/0 : 10[2] -> 11[3] via P2P/CUMEM cnode7-013:736483:745475 [2] NCCL INFO Channel 12/0 : 10[2] -> 11[3] via P2P/CUMEM cnode7-013:736483:745475 [2] NCCL INFO Channel 14/0 : 10[2] -> 11[3] via P2P/CUMEM cnode7-012:1272008:1281624 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/CUMEM cnode7-012:1272008:1281624 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/CUMEM cnode7-012:1272008:1281624 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/CUMEM cnode7-012:1272008:1281624 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/CUMEM cnode7-012:1272008:1281624 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/CUMEM cnode7-012:1272008:1281624 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/CUMEM cnode7-012:1272008:1281624 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/CUMEM cnode7-012:1272008:1281624 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/CUMEM cnode7-012:1272007:1281061 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/CUMEM cnode7-012:1272007:1281061 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/CUMEM cnode7-012:1272007:1281061 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/CUMEM cnode7-012:1272007:1281061 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/CUMEM cnode7-012:1272007:1281061 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/CUMEM cnode7-012:1272007:1281061 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/CUMEM cnode7-013:736484:745381 [3] NCCL INFO Channel 00/0 : 11[3] -> 12[4] via P2P/CUMEM cnode7-013:736484:745381 [3] NCCL INFO Channel 02/0 : 11[3] -> 12[4] via P2P/CUMEM cnode7-013:736484:745381 [3] NCCL INFO Channel 04/0 : 11[3] -> 12[4] via P2P/CUMEM cnode7-013:736484:745381 [3] NCCL INFO Channel 06/0 : 11[3] -> 12[4] via P2P/CUMEM cnode7-013:736484:745381 [3] NCCL INFO Channel 08/0 : 11[3] -> 12[4] via P2P/CUMEM cnode7-013:736484:745381 [3] NCCL INFO Channel 10/0 : 11[3] -> 12[4] via P2P/CUMEM cnode7-013:736484:745381 [3] NCCL INFO Channel 12/0 : 11[3] -> 12[4] via P2P/CUMEM cnode7-013:736484:745381 [3] NCCL INFO Channel 14/0 : 11[3] -> 12[4] via P2P/CUMEM cnode7-013:736487:746330 [6] NCCL INFO Channel 00/0 : 14[6] -> 15[7] via P2P/CUMEM cnode7-013:736487:746330 [6] NCCL INFO Channel 02/0 : 14[6] -> 15[7] via P2P/CUMEM cnode7-013:736487:746330 [6] NCCL INFO Channel 04/0 : 14[6] -> 15[7] via P2P/CUMEM cnode7-013:736487:746330 [6] NCCL INFO Channel 08/0 : 14[6] -> 15[7] via P2P/CUMEM cnode7-013:736487:746330 [6] NCCL INFO Channel 10/0 : 14[6] -> 15[7] via P2P/CUMEM cnode7-013:736487:746330 [6] NCCL INFO Channel 12/0 : 14[6] -> 15[7] via P2P/CUMEM cnode7-013:736487:746330 [6] NCCL INFO Channel 07/0 : 6[6] -> 14[6] [receive] via NET/IB/7/GDRDMA cnode7-013:736487:746330 [6] NCCL INFO Channel 15/0 : 6[6] -> 14[6] [receive] via NET/IB/7/GDRDMA cnode7-013:736487:746330 [6] NCCL INFO Channel 06/0 : 14[6] -> 6[6] [send] via NET/IB/7/GDRDMA cnode7-013:736487:746330 [6] NCCL INFO Channel 14/0 : 14[6] -> 6[6] [send] via NET/IB/7/GDRDMA cnode7-013:736487:746330 [6] NCCL INFO Channel 01/0 : 14[6] -> 13[5] via P2P/CUMEM cnode7-013:736487:746330 [6] NCCL INFO Channel 03/0 : 14[6] -> 13[5] via P2P/CUMEM cnode7-013:736487:746330 [6] NCCL INFO Channel 05/0 : 14[6] -> 13[5] via P2P/CUMEM cnode7-013:736487:746330 [6] NCCL INFO Channel 07/0 : 14[6] -> 13[5] via P2P/CUMEM cnode7-013:736487:746330 [6] NCCL INFO Channel 09/0 : 14[6] -> 13[5] via P2P/CUMEM cnode7-013:736487:746330 [6] NCCL INFO Channel 11/0 : 14[6] -> 13[5] via P2P/CUMEM cnode7-013:736487:746330 [6] NCCL INFO Channel 13/0 : 14[6] -> 13[5] via P2P/CUMEM cnode7-013:736487:746330 [6] NCCL INFO Channel 15/0 : 14[6] -> 13[5] via P2P/CUMEM cnode7-013:736481:744977 [0] NCCL INFO Channel 02/0 : 8[0] -> 9[1] via P2P/CUMEM cnode7-013:736481:744977 [0] NCCL INFO Channel 04/0 : 8[0] -> 9[1] via P2P/CUMEM cnode7-013:736481:744977 [0] NCCL INFO Channel 06/0 : 8[0] -> 9[1] via P2P/CUMEM cnode7-013:736481:744977 [0] NCCL INFO Channel 10/0 : 8[0] -> 9[1] via P2P/CUMEM cnode7-013:736481:744977 [0] NCCL INFO Channel 12/0 : 8[0] -> 9[1] via P2P/CUMEM cnode7-013:736481:744977 [0] NCCL INFO Channel 14/0 : 8[0] -> 9[1] via P2P/CUMEM cnode7-013:736485:745370 [4] NCCL INFO Channel 00/0 : 12[4] -> 13[5] via P2P/CUMEM cnode7-013:736485:745370 [4] NCCL INFO Channel 02/0 : 12[4] -> 13[5] via P2P/CUMEM cnode7-013:736485:745370 [4] NCCL INFO Channel 06/0 : 12[4] -> 13[5] via P2P/CUMEM cnode7-013:736485:745370 [4] NCCL INFO Channel 08/0 : 12[4] -> 13[5] via P2P/CUMEM cnode7-013:736485:745370 [4] NCCL INFO Channel 10/0 : 12[4] -> 13[5] via P2P/CUMEM cnode7-013:736485:745370 [4] NCCL INFO Channel 14/0 : 12[4] -> 13[5] via P2P/CUMEM cnode7-013:736486:745644 [5] NCCL INFO Channel 00/0 : 13[5] -> 14[6] via P2P/CUMEM cnode7-013:736486:745644 [5] NCCL INFO Channel 02/0 : 13[5] -> 14[6] via P2P/CUMEM cnode7-013:736486:745644 [5] NCCL INFO Channel 04/0 : 13[5] -> 14[6] via P2P/CUMEM cnode7-013:736486:745644 [5] NCCL INFO Channel 06/0 : 13[5] -> 14[6] via P2P/CUMEM cnode7-013:736486:745644 [5] NCCL INFO Channel 08/0 : 13[5] -> 14[6] via P2P/CUMEM cnode7-013:736486:745644 [5] NCCL INFO Channel 10/0 : 13[5] -> 14[6] via P2P/CUMEM cnode7-013:736486:745644 [5] NCCL INFO Channel 12/0 : 13[5] -> 14[6] via P2P/CUMEM cnode7-013:736486:745644 [5] NCCL INFO Channel 14/0 : 13[5] -> 14[6] via P2P/CUMEM cnode7-013:736486:745644 [5] NCCL INFO Channel 04/0 : 5[5] -> 13[5] [receive] via NET/IB/6/GDRDMA cnode7-013:736486:745644 [5] NCCL INFO Channel 12/0 : 5[5] -> 13[5] [receive] via NET/IB/6/GDRDMA cnode7-013:736486:745644 [5] NCCL INFO Channel 05/0 : 13[5] -> 5[5] [send] via NET/IB/6/GDRDMA cnode7-013:736486:745644 [5] NCCL INFO Channel 13/0 : 13[5] -> 5[5] [send] via NET/IB/6/GDRDMA cnode7-012:1272012:1281554 [7] NCCL INFO Channel 07/0 : 15[7] -> 7[7] [receive] via NET/IB/8/GDRDMA cnode7-012:1272012:1281554 [7] NCCL INFO Channel 15/0 : 15[7] -> 7[7] [receive] via NET/IB/8/GDRDMA cnode7-012:1272012:1281554 [7] NCCL INFO Channel 06/0 : 7[7] -> 15[7] [send] via NET/IB/8/GDRDMA cnode7-012:1272012:1281554 [7] NCCL INFO Channel 14/0 : 7[7] -> 15[7] [send] via NET/IB/8/GDRDMA cnode7-012:1272006:1281620 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/CUMEM cnode7-012:1272006:1281620 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/CUMEM cnode7-012:1272006:1281620 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/CUMEM cnode7-012:1272006:1281620 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/CUMEM cnode7-012:1272006:1281620 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/CUMEM cnode7-012:1272006:1281620 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/CUMEM cnode7-012:1272006:1281620 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/CUMEM cnode7-012:1272006:1281620 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/CUMEM cnode7-012:1272006:1281620 [1] NCCL INFO Channel 01/0 : 9[1] -> 1[1] [receive] via NET/IB/1/GDRDMA cnode7-012:1272006:1281620 [1] NCCL INFO Channel 09/0 : 9[1] -> 1[1] [receive] via NET/IB/1/GDRDMA cnode7-012:1272006:1281620 [1] NCCL INFO Channel 00/0 : 1[1] -> 9[1] [send] via NET/IB/1/GDRDMA cnode7-012:1272006:1281620 [1] NCCL INFO Channel 08/0 : 1[1] -> 9[1] [send] via NET/IB/1/GDRDMA cnode7-012:1272006:1281620 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/CUMEM cnode7-012:1272006:1281620 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/CUMEM cnode7-012:1272006:1281620 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/CUMEM cnode7-012:1272011:1281645 [6] NCCL INFO Channel 06/0 : 14[6] -> 6[6] [receive] via NET/IB/7/GDRDMA cnode7-012:1272011:1281645 [6] NCCL INFO Channel 14/0 : 14[6] -> 6[6] [receive] via NET/IB/7/GDRDMA cnode7-012:1272011:1281645 [6] NCCL INFO Channel 07/0 : 6[6] -> 14[6] [send] via NET/IB/7/GDRDMA cnode7-012:1272011:1281645 [6] NCCL INFO Channel 15/0 : 6[6] -> 14[6] [send] via NET/IB/7/GDRDMA cnode7-012:1272011:1281645 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/CUMEM cnode7-012:1272011:1281645 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/CUMEM cnode7-012:1272011:1281645 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/CUMEM cnode7-012:1272011:1281645 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/CUMEM cnode7-012:1272011:1281645 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/CUMEM cnode7-012:1272011:1281645 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/CUMEM cnode7-012:1272011:1281645 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/CUMEM cnode7-012:1272011:1281645 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/CUMEM cnode7-013:736483:745475 [2] NCCL INFO Channel 03/0 : 2[2] -> 10[2] [receive] via NET/IB/2/GDRDMA cnode7-013:736483:745475 [2] NCCL INFO Channel 11/0 : 2[2] -> 10[2] [receive] via NET/IB/2/GDRDMA cnode7-013:736483:745475 [2] NCCL INFO Channel 02/0 : 10[2] -> 2[2] [send] via NET/IB/2/GDRDMA cnode7-013:736483:745475 [2] NCCL INFO Channel 10/0 : 10[2] -> 2[2] [send] via NET/IB/2/GDRDMA cnode7-013:736483:745475 [2] NCCL INFO Channel 01/0 : 10[2] -> 9[1] via P2P/CUMEM cnode7-013:736483:745475 [2] NCCL INFO Channel 03/0 : 10[2] -> 9[1] via P2P/CUMEM cnode7-013:736483:745475 [2] NCCL INFO Channel 05/0 : 10[2] -> 9[1] via P2P/CUMEM cnode7-013:736483:745475 [2] NCCL INFO Channel 07/0 : 10[2] -> 9[1] via P2P/CUMEM cnode7-013:736483:745475 [2] NCCL INFO Channel 09/0 : 10[2] -> 9[1] via P2P/CUMEM cnode7-013:736483:745475 [2] NCCL INFO Channel 11/0 : 10[2] -> 9[1] via P2P/CUMEM cnode7-013:736483:745475 [2] NCCL INFO Channel 13/0 : 10[2] -> 9[1] via P2P/CUMEM cnode7-013:736483:745475 [2] NCCL INFO Channel 15/0 : 10[2] -> 9[1] via P2P/CUMEM cnode7-012:1272010:1281060 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/CUMEM cnode7-012:1272010:1281060 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/CUMEM cnode7-012:1272010:1281060 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/CUMEM cnode7-012:1272010:1281060 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/CUMEM cnode7-012:1272010:1281060 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/CUMEM cnode7-012:1272010:1281060 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/CUMEM cnode7-012:1272010:1281060 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/CUMEM cnode7-012:1272010:1281060 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/CUMEM cnode7-012:1272010:1281060 [5] NCCL INFO Channel 05/0 : 13[5] -> 5[5] [receive] via NET/IB/6/GDRDMA cnode7-012:1272010:1281060 [5] NCCL INFO Channel 13/0 : 13[5] -> 5[5] [receive] via NET/IB/6/GDRDMA cnode7-012:1272010:1281060 [5] NCCL INFO Channel 04/0 : 5[5] -> 13[5] [send] via NET/IB/6/GDRDMA cnode7-012:1272010:1281060 [5] NCCL INFO Channel 12/0 : 5[5] -> 13[5] [send] via NET/IB/6/GDRDMA cnode7-012:1272005:1281059 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/CUMEM cnode7-012:1272005:1281059 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/CUMEM cnode7-012:1272005:1281059 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/CUMEM cnode7-012:1272005:1281059 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/CUMEM cnode7-012:1272005:1281059 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/CUMEM cnode7-012:1272005:1281059 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/CUMEM cnode7-012:1272005:1281059 [0] NCCL INFO Channel 00/0 : 0[0] -> 7[7] via P2P/CUMEM cnode7-012:1272005:1281059 [0] NCCL INFO Channel 02/0 : 0[0] -> 7[7] via P2P/CUMEM cnode7-012:1272005:1281059 [0] NCCL INFO Channel 04/0 : 0[0] -> 7[7] via P2P/CUMEM cnode7-012:1272005:1281059 [0] NCCL INFO Channel 06/0 : 0[0] -> 7[7] via P2P/CUMEM cnode7-012:1272005:1281059 [0] NCCL INFO Channel 08/0 : 0[0] -> 7[7] via P2P/CUMEM cnode7-012:1272005:1281059 [0] NCCL INFO Channel 10/0 : 0[0] -> 7[7] via P2P/CUMEM cnode7-012:1272005:1281059 [0] NCCL INFO Channel 12/0 : 0[0] -> 7[7] via P2P/CUMEM cnode7-012:1272005:1281059 [0] NCCL INFO Channel 14/0 : 0[0] -> 7[7] via P2P/CUMEM cnode7-012:1272005:1281059 [0] NCCL INFO Channel 00/0 : 8[0] -> 0[0] [receive] via NET/IB/0/GDRDMA cnode7-012:1272005:1281059 [0] NCCL INFO Channel 08/0 : 8[0] -> 0[0] [receive] via NET/IB/0/GDRDMA cnode7-012:1272005:1281059 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [send] via NET/IB/0/GDRDMA cnode7-012:1272005:1281059 [0] NCCL INFO Channel 09/0 : 0[0] -> 8[0] [send] via NET/IB/0/GDRDMA cnode7-013:736488:746041 [7] NCCL INFO Channel 06/0 : 7[7] -> 15[7] [receive] via NET/IB/8/GDRDMA cnode7-013:736488:746041 [7] NCCL INFO Channel 14/0 : 7[7] -> 15[7] [receive] via NET/IB/8/GDRDMA cnode7-013:736488:746041 [7] NCCL INFO Channel 07/0 : 15[7] -> 7[7] [send] via NET/IB/8/GDRDMA cnode7-013:736488:746041 [7] NCCL INFO Channel 15/0 : 15[7] -> 7[7] [send] via NET/IB/8/GDRDMA cnode7-013:736488:746041 [7] NCCL INFO Channel 00/0 : 15[7] -> 8[0] via P2P/CUMEM cnode7-013:736488:746041 [7] NCCL INFO Channel 02/0 : 15[7] -> 8[0] via P2P/CUMEM cnode7-013:736488:746041 [7] NCCL INFO Channel 04/0 : 15[7] -> 8[0] via P2P/CUMEM cnode7-013:736488:746041 [7] NCCL INFO Channel 06/0 : 15[7] -> 8[0] via P2P/CUMEM cnode7-013:736488:746041 [7] NCCL INFO Channel 08/0 : 15[7] -> 8[0] via P2P/CUMEM cnode7-013:736488:746041 [7] NCCL INFO Channel 10/0 : 15[7] -> 8[0] via P2P/CUMEM cnode7-013:736488:746041 [7] NCCL INFO Channel 12/0 : 15[7] -> 8[0] via P2P/CUMEM cnode7-013:736488:746041 [7] NCCL INFO Channel 14/0 : 15[7] -> 8[0] via P2P/CUMEM cnode7-013:736488:746041 [7] NCCL INFO Channel 01/0 : 15[7] -> 14[6] via P2P/CUMEM cnode7-013:736488:746041 [7] NCCL INFO Channel 03/0 : 15[7] -> 14[6] via P2P/CUMEM cnode7-013:736488:746041 [7] NCCL INFO Channel 05/0 : 15[7] -> 14[6] via P2P/CUMEM cnode7-013:736488:746041 [7] NCCL INFO Channel 09/0 : 15[7] -> 14[6] via P2P/CUMEM cnode7-013:736488:746041 [7] NCCL INFO Channel 11/0 : 15[7] -> 14[6] via P2P/CUMEM cnode7-013:736488:746041 [7] NCCL INFO Channel 13/0 : 15[7] -> 14[6] via P2P/CUMEM cnode7-012:1272008:1281624 [3] NCCL INFO Channel 03/0 : 11[3] -> 3[3] [receive] via NET/IB/3/GDRDMA cnode7-012:1272008:1281624 [3] NCCL INFO Channel 11/0 : 11[3] -> 3[3] [receive] via NET/IB/3/GDRDMA cnode7-012:1272008:1281624 [3] NCCL INFO Channel 02/0 : 3[3] -> 11[3] [send] via NET/IB/3/GDRDMA cnode7-012:1272008:1281624 [3] NCCL INFO Channel 10/0 : 3[3] -> 11[3] [send] via NET/IB/3/GDRDMA cnode7-013:736482:746397 [1] NCCL INFO Channel 00/0 : 9[1] -> 10[2] via P2P/CUMEM cnode7-013:736482:746397 [1] NCCL INFO Channel 02/0 : 9[1] -> 10[2] via P2P/CUMEM cnode7-013:736482:746397 [1] NCCL INFO Channel 04/0 : 9[1] -> 10[2] via P2P/CUMEM cnode7-013:736482:746397 [1] NCCL INFO Channel 06/0 : 9[1] -> 10[2] via P2P/CUMEM cnode7-013:736482:746397 [1] NCCL INFO Channel 08/0 : 9[1] -> 10[2] via P2P/CUMEM cnode7-013:736482:746397 [1] NCCL INFO Channel 10/0 : 9[1] -> 10[2] via P2P/CUMEM cnode7-013:736482:746397 [1] NCCL INFO Channel 12/0 : 9[1] -> 10[2] via P2P/CUMEM cnode7-013:736482:746397 [1] NCCL INFO Channel 14/0 : 9[1] -> 10[2] via P2P/CUMEM cnode7-013:736482:746397 [1] NCCL INFO Channel 00/0 : 1[1] -> 9[1] [receive] via NET/IB/1/GDRDMA cnode7-013:736482:746397 [1] NCCL INFO Channel 08/0 : 1[1] -> 9[1] [receive] via NET/IB/1/GDRDMA cnode7-013:736482:746397 [1] NCCL INFO Channel 01/0 : 9[1] -> 1[1] [send] via NET/IB/1/GDRDMA cnode7-013:736482:746397 [1] NCCL INFO Channel 09/0 : 9[1] -> 1[1] [send] via NET/IB/1/GDRDMA cnode7-013:736482:746397 [1] NCCL INFO Channel 03/0 : 9[1] -> 8[0] via P2P/CUMEM cnode7-013:736482:746397 [1] NCCL INFO Channel 05/0 : 9[1] -> 8[0] via P2P/CUMEM cnode7-013:736482:746397 [1] NCCL INFO Channel 07/0 : 9[1] -> 8[0] via P2P/CUMEM cnode7-013:736482:746397 [1] NCCL INFO Channel 11/0 : 9[1] -> 8[0] via P2P/CUMEM cnode7-013:736482:746397 [1] NCCL INFO Channel 13/0 : 9[1] -> 8[0] via P2P/CUMEM cnode7-013:736482:746397 [1] NCCL INFO Channel 15/0 : 9[1] -> 8[0] via P2P/CUMEM cnode7-012:1272007:1281061 [2] NCCL INFO Channel 02/0 : 10[2] -> 2[2] [receive] via NET/IB/2/GDRDMA cnode7-012:1272007:1281061 [2] NCCL INFO Channel 10/0 : 10[2] -> 2[2] [receive] via NET/IB/2/GDRDMA cnode7-012:1272007:1281061 [2] NCCL INFO Channel 03/0 : 2[2] -> 10[2] [send] via NET/IB/2/GDRDMA cnode7-012:1272007:1281061 [2] NCCL INFO Channel 11/0 : 2[2] -> 10[2] [send] via NET/IB/2/GDRDMA cnode7-012:1272007:1281061 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/CUMEM cnode7-012:1272007:1281061 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/CUMEM cnode7-012:1272007:1281061 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/CUMEM cnode7-012:1272007:1281061 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/CUMEM cnode7-012:1272007:1281061 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/CUMEM cnode7-012:1272007:1281061 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/CUMEM cnode7-012:1272007:1281061 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/CUMEM cnode7-012:1272007:1281061 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/CUMEM cnode7-013:736484:745381 [3] NCCL INFO Channel 02/0 : 3[3] -> 11[3] [receive] via NET/IB/3/GDRDMA cnode7-013:736484:745381 [3] NCCL INFO Channel 10/0 : 3[3] -> 11[3] [receive] via NET/IB/3/GDRDMA cnode7-013:736484:745381 [3] NCCL INFO Channel 03/0 : 11[3] -> 3[3] [send] via NET/IB/3/GDRDMA cnode7-013:736484:745381 [3] NCCL INFO Channel 11/0 : 11[3] -> 3[3] [send] via NET/IB/3/GDRDMA cnode7-013:736484:745381 [3] NCCL INFO Channel 01/0 : 11[3] -> 10[2] via P2P/CUMEM cnode7-013:736484:745381 [3] NCCL INFO Channel 05/0 : 11[3] -> 10[2] via P2P/CUMEM cnode7-013:736484:745381 [3] NCCL INFO Channel 07/0 : 11[3] -> 10[2] via P2P/CUMEM cnode7-013:736484:745381 [3] NCCL INFO Channel 09/0 : 11[3] -> 10[2] via P2P/CUMEM cnode7-013:736484:745381 [3] NCCL INFO Channel 13/0 : 11[3] -> 10[2] via P2P/CUMEM cnode7-013:736481:744977 [0] NCCL INFO Channel 01/0 : 8[0] -> 15[7] via P2P/CUMEM cnode7-013:736481:744977 [0] NCCL INFO Channel 03/0 : 8[0] -> 15[7] via P2P/CUMEM cnode7-013:736481:744977 [0] NCCL INFO Channel 05/0 : 8[0] -> 15[7] via P2P/CUMEM cnode7-013:736481:744977 [0] NCCL INFO Channel 07/0 : 8[0] -> 15[7] via P2P/CUMEM cnode7-013:736481:744977 [0] NCCL INFO Channel 09/0 : 8[0] -> 15[7] via P2P/CUMEM cnode7-013:736481:744977 [0] NCCL INFO Channel 11/0 : 8[0] -> 15[7] via P2P/CUMEM cnode7-013:736481:744977 [0] NCCL INFO Channel 13/0 : 8[0] -> 15[7] via P2P/CUMEM cnode7-013:736481:744977 [0] NCCL INFO Channel 15/0 : 8[0] -> 15[7] via P2P/CUMEM cnode7-013:736481:744977 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [receive] via NET/IB/0/GDRDMA cnode7-013:736481:744977 [0] NCCL INFO Channel 09/0 : 0[0] -> 8[0] [receive] via NET/IB/0/GDRDMA cnode7-013:736481:744977 [0] NCCL INFO Channel 00/0 : 8[0] -> 0[0] [send] via NET/IB/0/GDRDMA cnode7-013:736481:744977 [0] NCCL INFO Channel 08/0 : 8[0] -> 0[0] [send] via NET/IB/0/GDRDMA cnode7-013:736485:745370 [4] NCCL INFO Channel 05/0 : 4[4] -> 12[4] [receive] via NET/IB/4/GDRDMA cnode7-013:736485:745370 [4] NCCL INFO Channel 13/0 : 4[4] -> 12[4] [receive] via NET/IB/4/GDRDMA cnode7-013:736485:745370 [4] NCCL INFO Channel 04/0 : 12[4] -> 4[4] [send] via NET/IB/4/GDRDMA cnode7-013:736485:745370 [4] NCCL INFO Channel 12/0 : 12[4] -> 4[4] [send] via NET/IB/4/GDRDMA cnode7-013:736485:745370 [4] NCCL INFO Channel 01/0 : 12[4] -> 11[3] via P2P/CUMEM cnode7-013:736485:745370 [4] NCCL INFO Channel 03/0 : 12[4] -> 11[3] via P2P/CUMEM cnode7-013:736485:745370 [4] NCCL INFO Channel 05/0 : 12[4] -> 11[3] via P2P/CUMEM cnode7-013:736485:745370 [4] NCCL INFO Channel 07/0 : 12[4] -> 11[3] via P2P/CUMEM cnode7-013:736485:745370 [4] NCCL INFO Channel 09/0 : 12[4] -> 11[3] via P2P/CUMEM cnode7-013:736485:745370 [4] NCCL INFO Channel 11/0 : 12[4] -> 11[3] via P2P/CUMEM cnode7-013:736485:745370 [4] NCCL INFO Channel 13/0 : 12[4] -> 11[3] via P2P/CUMEM cnode7-013:736485:745370 [4] NCCL INFO Channel 15/0 : 12[4] -> 11[3] via P2P/CUMEM cnode7-013:736486:745644 [5] NCCL INFO Channel 01/0 : 13[5] -> 12[4] via P2P/CUMEM cnode7-013:736486:745644 [5] NCCL INFO Channel 03/0 : 13[5] -> 12[4] via P2P/CUMEM cnode7-013:736486:745644 [5] NCCL INFO Channel 07/0 : 13[5] -> 12[4] via P2P/CUMEM cnode7-013:736486:745644 [5] NCCL INFO Channel 09/0 : 13[5] -> 12[4] via P2P/CUMEM cnode7-013:736486:745644 [5] NCCL INFO Channel 11/0 : 13[5] -> 12[4] via P2P/CUMEM cnode7-013:736486:745644 [5] NCCL INFO Channel 15/0 : 13[5] -> 12[4] via P2P/CUMEM cnode7-012:1272012:1281554 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/CUMEM cnode7-012:1272012:1281554 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/CUMEM cnode7-012:1272012:1281554 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/CUMEM cnode7-012:1272012:1281554 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/CUMEM cnode7-012:1272012:1281554 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/CUMEM cnode7-012:1272012:1281554 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/CUMEM cnode7-012:1272012:1281554 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/CUMEM cnode7-012:1272012:1281554 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/CUMEM cnode7-012:1272012:1281554 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/CUMEM cnode7-012:1272012:1281554 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/CUMEM cnode7-012:1272012:1281554 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/CUMEM cnode7-012:1272012:1281554 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/CUMEM cnode7-012:1272012:1281554 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/CUMEM cnode7-012:1272012:1281554 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/CUMEM cnode7-012:1272009:1281281 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/CUMEM cnode7-012:1272009:1281281 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/CUMEM cnode7-012:1272009:1281281 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/CUMEM cnode7-012:1272009:1281281 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/CUMEM cnode7-012:1272009:1281281 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/CUMEM cnode7-012:1272009:1281281 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/CUMEM cnode7-012:1272009:1281281 [4] NCCL INFO Channel 04/0 : 12[4] -> 4[4] [receive] via NET/IB/4/GDRDMA cnode7-012:1272009:1281281 [4] NCCL INFO Channel 12/0 : 12[4] -> 4[4] [receive] via NET/IB/4/GDRDMA cnode7-012:1272009:1281281 [4] NCCL INFO Channel 05/0 : 4[4] -> 12[4] [send] via NET/IB/4/GDRDMA cnode7-012:1272009:1281281 [4] NCCL INFO Channel 13/0 : 4[4] -> 12[4] [send] via NET/IB/4/GDRDMA cnode7-012:1272009:1281281 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/CUMEM cnode7-012:1272009:1281281 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/CUMEM cnode7-012:1272009:1281281 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/CUMEM cnode7-012:1272009:1281281 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/CUMEM cnode7-012:1272009:1281281 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/CUMEM cnode7-012:1272009:1281281 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/CUMEM cnode7-012:1272009:1281281 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/CUMEM cnode7-012:1272009:1281281 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/CUMEM cnode7-012:1272006:1281620 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/CUMEM cnode7-012:1272006:1281620 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/CUMEM cnode7-012:1272006:1281620 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/CUMEM cnode7-012:1272010:1281060 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/CUMEM cnode7-012:1272010:1281060 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/CUMEM cnode7-012:1272010:1281060 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/CUMEM cnode7-012:1272010:1281060 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/CUMEM cnode7-012:1272010:1281060 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/CUMEM cnode7-012:1272010:1281060 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/CUMEM cnode7-012:1272008:1281624 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/CUMEM cnode7-012:1272008:1281624 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/CUMEM cnode7-012:1272008:1281624 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/CUMEM cnode7-012:1272008:1281624 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/CUMEM cnode7-012:1272008:1281624 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/CUMEM cnode7-012:1272008:1281624 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/CUMEM cnode7-013:736484:745381 [3] NCCL INFO Channel 15/0 : 11[3] -> 10[2] via P2P/CUMEM cnode7-012:1272007:1281061 [2] NCCL INFO Connected all rings cnode7-013:736481:744977 [0] NCCL INFO Connected all rings cnode7-013:736481:744977 [0] NCCL INFO Channel 00/0 : 8[0] -> 9[1] via P2P/CUMEM cnode7-013:736481:744977 [0] NCCL INFO Channel 03/0 : 8[0] -> 9[1] via P2P/CUMEM cnode7-012:1272009:1281281 [4] NCCL INFO Connected all rings cnode7-012:1272009:1281281 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/CUMEM cnode7-012:1272009:1281281 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/CUMEM cnode7-012:1272009:1281281 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/CUMEM cnode7-012:1272009:1281281 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/CUMEM cnode7-012:1272009:1281281 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/CUMEM cnode7-012:1272009:1281281 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/CUMEM cnode7-012:1272009:1281281 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/CUMEM cnode7-012:1272006:1281620 [1] NCCL INFO Connected all rings cnode7-012:1272006:1281620 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/CUMEM cnode7-012:1272006:1281620 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/CUMEM cnode7-012:1272006:1281620 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/CUMEM cnode7-012:1272006:1281620 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/CUMEM cnode7-012:1272006:1281620 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/CUMEM cnode7-012:1272006:1281620 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/CUMEM cnode7-012:1272006:1281620 [1] NCCL INFO Channel 01/0 : 1[1] -> 9[1] [send] via NET/IB/1/GDRDMA cnode7-012:1272006:1281620 [1] NCCL INFO Channel 09/0 : 1[1] -> 9[1] [send] via NET/IB/1/GDRDMA cnode7-012:1272006:1281620 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/CUMEM cnode7-012:1272006:1281620 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/CUMEM cnode7-012:1272006:1281620 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/CUMEM cnode7-012:1272006:1281620 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/CUMEM cnode7-012:1272006:1281620 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/CUMEM cnode7-012:1272006:1281620 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/CUMEM cnode7-012:1272006:1281620 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/CUMEM cnode7-012:1272006:1281620 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/CUMEM cnode7-012:1272011:1281645 [6] NCCL INFO Connected all rings cnode7-012:1272011:1281645 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/CUMEM cnode7-012:1272011:1281645 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/CUMEM cnode7-012:1272011:1281645 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/CUMEM cnode7-012:1272011:1281645 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/CUMEM cnode7-012:1272011:1281645 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/CUMEM cnode7-012:1272011:1281645 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/CUMEM cnode7-012:1272011:1281645 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/CUMEM cnode7-012:1272011:1281645 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/CUMEM cnode7-012:1272011:1281645 [6] NCCL INFO Channel 06/0 : 6[6] -> 14[6] [send] via NET/IB/7/GDRDMA cnode7-012:1272011:1281645 [6] NCCL INFO Channel 14/0 : 6[6] -> 14[6] [send] via NET/IB/7/GDRDMA cnode7-013:736483:745475 [2] NCCL INFO Connected all rings cnode7-013:736483:745475 [2] NCCL INFO Channel 01/0 : 10[2] -> 11[3] via P2P/CUMEM cnode7-013:736483:745475 [2] NCCL INFO Channel 02/0 : 10[2] -> 11[3] via P2P/CUMEM cnode7-013:736483:745475 [2] NCCL INFO Channel 05/0 : 10[2] -> 11[3] via P2P/CUMEM cnode7-013:736483:745475 [2] NCCL INFO Channel 07/0 : 10[2] -> 11[3] via P2P/CUMEM cnode7-013:736483:745475 [2] NCCL INFO Channel 09/0 : 10[2] -> 11[3] via P2P/CUMEM cnode7-013:736483:745475 [2] NCCL INFO Channel 10/0 : 10[2] -> 11[3] via P2P/CUMEM cnode7-013:736483:745475 [2] NCCL INFO Channel 13/0 : 10[2] -> 11[3] via P2P/CUMEM cnode7-013:736483:745475 [2] NCCL INFO Channel 15/0 : 10[2] -> 11[3] via P2P/CUMEM cnode7-013:736483:745475 [2] NCCL INFO Channel 02/0 : 2[2] -> 10[2] [receive] via NET/IB/2/GDRDMA cnode7-013:736483:745475 [2] NCCL INFO Channel 10/0 : 2[2] -> 10[2] [receive] via NET/IB/2/GDRDMA cnode7-013:736483:745475 [2] NCCL INFO Channel 00/0 : 10[2] -> 9[1] via P2P/CUMEM cnode7-013:736483:745475 [2] NCCL INFO Channel 04/0 : 10[2] -> 9[1] via P2P/CUMEM cnode7-013:736483:745475 [2] NCCL INFO Channel 06/0 : 10[2] -> 9[1] via P2P/CUMEM cnode7-013:736483:745475 [2] NCCL INFO Channel 08/0 : 10[2] -> 9[1] via P2P/CUMEM cnode7-013:736483:745475 [2] NCCL INFO Channel 12/0 : 10[2] -> 9[1] via P2P/CUMEM cnode7-013:736483:745475 [2] NCCL INFO Channel 14/0 : 10[2] -> 9[1] via P2P/CUMEM cnode7-012:1272010:1281060 [5] NCCL INFO Connected all rings cnode7-012:1272010:1281060 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/CUMEM cnode7-012:1272010:1281060 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/CUMEM cnode7-012:1272010:1281060 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/CUMEM cnode7-012:1272010:1281060 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/CUMEM cnode7-012:1272010:1281060 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/CUMEM cnode7-012:1272010:1281060 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/CUMEM cnode7-012:1272010:1281060 [5] NCCL INFO Channel 05/0 : 5[5] -> 13[5] [send] via NET/IB/6/GDRDMA cnode7-012:1272010:1281060 [5] NCCL INFO Channel 13/0 : 5[5] -> 13[5] [send] via NET/IB/6/GDRDMA cnode7-012:1272005:1281059 [0] NCCL INFO Connected all rings cnode7-012:1272005:1281059 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/CUMEM cnode7-012:1272005:1281059 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/CUMEM cnode7-012:1272005:1281059 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/CUMEM cnode7-012:1272005:1281059 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/CUMEM cnode7-012:1272005:1281059 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/CUMEM cnode7-012:1272005:1281059 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/CUMEM cnode7-012:1272005:1281059 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/CUMEM cnode7-012:1272005:1281059 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/CUMEM cnode7-012:1272005:1281059 [0] NCCL INFO Channel 01/0 : 0[0] -> 7[7] via P2P/CUMEM cnode7-012:1272005:1281059 [0] NCCL INFO Channel 03/0 : 0[0] -> 7[7] via P2P/CUMEM cnode7-012:1272005:1281059 [0] NCCL INFO Channel 05/0 : 0[0] -> 7[7] via P2P/CUMEM cnode7-012:1272005:1281059 [0] NCCL INFO Channel 07/0 : 0[0] -> 7[7] via P2P/CUMEM cnode7-012:1272005:1281059 [0] NCCL INFO Channel 09/0 : 0[0] -> 7[7] via P2P/CUMEM cnode7-012:1272005:1281059 [0] NCCL INFO Channel 11/0 : 0[0] -> 7[7] via P2P/CUMEM cnode7-012:1272005:1281059 [0] NCCL INFO Channel 13/0 : 0[0] -> 7[7] via P2P/CUMEM cnode7-012:1272005:1281059 [0] NCCL INFO Channel 15/0 : 0[0] -> 7[7] via P2P/CUMEM cnode7-012:1272005:1281059 [0] NCCL INFO Channel 00/0 : 0[0] -> 8[0] [send] via NET/IB/0/GDRDMA cnode7-012:1272005:1281059 [0] NCCL INFO Channel 08/0 : 0[0] -> 8[0] [send] via NET/IB/0/GDRDMA cnode7-013:736488:746041 [7] NCCL INFO Connected all rings cnode7-013:736488:746041 [7] NCCL INFO Channel 07/0 : 7[7] -> 15[7] [receive] via NET/IB/8/GDRDMA cnode7-013:736488:746041 [7] NCCL INFO Channel 15/0 : 7[7] -> 15[7] [receive] via NET/IB/8/GDRDMA cnode7-013:736488:746041 [7] NCCL INFO Channel 01/0 : 15[7] -> 8[0] via P2P/CUMEM cnode7-013:736488:746041 [7] NCCL INFO Channel 03/0 : 15[7] -> 8[0] via P2P/CUMEM cnode7-013:736488:746041 [7] NCCL INFO Channel 05/0 : 15[7] -> 8[0] via P2P/CUMEM cnode7-013:736488:746041 [7] NCCL INFO Channel 07/0 : 15[7] -> 8[0] via P2P/CUMEM cnode7-013:736488:746041 [7] NCCL INFO Channel 09/0 : 15[7] -> 8[0] via P2P/CUMEM cnode7-013:736488:746041 [7] NCCL INFO Channel 11/0 : 15[7] -> 8[0] via P2P/CUMEM cnode7-013:736488:746041 [7] NCCL INFO Channel 13/0 : 15[7] -> 8[0] via P2P/CUMEM cnode7-013:736488:746041 [7] NCCL INFO Channel 15/0 : 15[7] -> 8[0] via P2P/CUMEM cnode7-013:736488:746041 [7] NCCL INFO Channel 00/0 : 15[7] -> 14[6] via P2P/CUMEM cnode7-013:736488:746041 [7] NCCL INFO Channel 02/0 : 15[7] -> 14[6] via P2P/CUMEM cnode7-013:736488:746041 [7] NCCL INFO Channel 04/0 : 15[7] -> 14[6] via P2P/CUMEM cnode7-013:736488:746041 [7] NCCL INFO Channel 06/0 : 15[7] -> 14[6] via P2P/CUMEM cnode7-013:736488:746041 [7] NCCL INFO Channel 08/0 : 15[7] -> 14[6] via P2P/CUMEM cnode7-013:736488:746041 [7] NCCL INFO Channel 10/0 : 15[7] -> 14[6] via P2P/CUMEM cnode7-013:736488:746041 [7] NCCL INFO Channel 12/0 : 15[7] -> 14[6] via P2P/CUMEM cnode7-012:1272008:1281624 [3] NCCL INFO Connected all rings cnode7-012:1272008:1281624 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/CUMEM cnode7-012:1272008:1281624 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/CUMEM cnode7-012:1272008:1281624 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/CUMEM cnode7-012:1272008:1281624 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/CUMEM cnode7-012:1272008:1281624 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/CUMEM cnode7-012:1272008:1281624 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/CUMEM cnode7-012:1272008:1281624 [3] NCCL INFO Channel 03/0 : 3[3] -> 11[3] [send] via NET/IB/3/GDRDMA cnode7-012:1272008:1281624 [3] NCCL INFO Channel 11/0 : 3[3] -> 11[3] [send] via NET/IB/3/GDRDMA cnode7-013:736482:746397 [1] NCCL INFO Connected all rings cnode7-013:736482:746397 [1] NCCL INFO Channel 01/0 : 9[1] -> 10[2] via P2P/CUMEM cnode7-013:736482:746397 [1] NCCL INFO Channel 03/0 : 9[1] -> 10[2] via P2P/CUMEM cnode7-013:736482:746397 [1] NCCL INFO Channel 05/0 : 9[1] -> 10[2] via P2P/CUMEM cnode7-013:736482:746397 [1] NCCL INFO Channel 07/0 : 9[1] -> 10[2] via P2P/CUMEM cnode7-013:736482:746397 [1] NCCL INFO Channel 09/0 : 9[1] -> 10[2] via P2P/CUMEM cnode7-013:736482:746397 [1] NCCL INFO Channel 11/0 : 9[1] -> 10[2] via P2P/CUMEM cnode7-013:736482:746397 [1] NCCL INFO Channel 13/0 : 9[1] -> 10[2] via P2P/CUMEM cnode7-013:736482:746397 [1] NCCL INFO Channel 15/0 : 9[1] -> 10[2] via P2P/CUMEM cnode7-013:736482:746397 [1] NCCL INFO Channel 01/0 : 1[1] -> 9[1] [receive] via NET/IB/1/GDRDMA cnode7-013:736482:746397 [1] NCCL INFO Channel 09/0 : 1[1] -> 9[1] [receive] via NET/IB/1/GDRDMA cnode7-013:736482:746397 [1] NCCL INFO Channel 00/0 : 9[1] -> 8[0] via P2P/CUMEM cnode7-013:736482:746397 [1] NCCL INFO Channel 02/0 : 9[1] -> 8[0] via P2P/CUMEM cnode7-013:736482:746397 [1] NCCL INFO Channel 04/0 : 9[1] -> 8[0] via P2P/CUMEM cnode7-013:736482:746397 [1] NCCL INFO Channel 06/0 : 9[1] -> 8[0] via P2P/CUMEM cnode7-013:736482:746397 [1] NCCL INFO Channel 08/0 : 9[1] -> 8[0] via P2P/CUMEM cnode7-013:736482:746397 [1] NCCL INFO Channel 10/0 : 9[1] -> 8[0] via P2P/CUMEM cnode7-013:736482:746397 [1] NCCL INFO Channel 12/0 : 9[1] -> 8[0] via P2P/CUMEM cnode7-013:736482:746397 [1] NCCL INFO Channel 14/0 : 9[1] -> 8[0] via P2P/CUMEM cnode7-012:1272007:1281061 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/CUMEM cnode7-012:1272007:1281061 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/CUMEM cnode7-012:1272007:1281061 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/CUMEM cnode7-012:1272007:1281061 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/CUMEM cnode7-012:1272007:1281061 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/CUMEM cnode7-012:1272007:1281061 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/CUMEM cnode7-012:1272007:1281061 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/CUMEM cnode7-012:1272007:1281061 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/CUMEM cnode7-012:1272007:1281061 [2] NCCL INFO Channel 02/0 : 2[2] -> 10[2] [send] via NET/IB/2/GDRDMA cnode7-012:1272007:1281061 [2] NCCL INFO Channel 10/0 : 2[2] -> 10[2] [send] via NET/IB/2/GDRDMA cnode7-012:1272007:1281061 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/CUMEM cnode7-012:1272007:1281061 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/CUMEM cnode7-012:1272007:1281061 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/CUMEM cnode7-012:1272007:1281061 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/CUMEM cnode7-012:1272007:1281061 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/CUMEM cnode7-012:1272007:1281061 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/CUMEM cnode7-013:736484:745381 [3] NCCL INFO Connected all rings cnode7-013:736484:745381 [3] NCCL INFO Channel 01/0 : 11[3] -> 12[4] via P2P/CUMEM cnode7-013:736484:745381 [3] NCCL INFO Channel 03/0 : 11[3] -> 12[4] via P2P/CUMEM cnode7-013:736484:745381 [3] NCCL INFO Channel 05/0 : 11[3] -> 12[4] via P2P/CUMEM cnode7-013:736484:745381 [3] NCCL INFO Channel 07/0 : 11[3] -> 12[4] via P2P/CUMEM cnode7-013:736484:745381 [3] NCCL INFO Channel 09/0 : 11[3] -> 12[4] via P2P/CUMEM cnode7-013:736484:745381 [3] NCCL INFO Channel 11/0 : 11[3] -> 12[4] via P2P/CUMEM cnode7-013:736484:745381 [3] NCCL INFO Channel 13/0 : 11[3] -> 12[4] via P2P/CUMEM cnode7-013:736484:745381 [3] NCCL INFO Channel 15/0 : 11[3] -> 12[4] via P2P/CUMEM cnode7-013:736484:745381 [3] NCCL INFO Channel 03/0 : 3[3] -> 11[3] [receive] via NET/IB/3/GDRDMA cnode7-013:736484:745381 [3] NCCL INFO Channel 11/0 : 3[3] -> 11[3] [receive] via NET/IB/3/GDRDMA cnode7-013:736484:745381 [3] NCCL INFO Channel 00/0 : 11[3] -> 10[2] via P2P/CUMEM cnode7-013:736484:745381 [3] NCCL INFO Channel 02/0 : 11[3] -> 10[2] via P2P/CUMEM cnode7-013:736484:745381 [3] NCCL INFO Channel 04/0 : 11[3] -> 10[2] via P2P/CUMEM cnode7-013:736484:745381 [3] NCCL INFO Channel 06/0 : 11[3] -> 10[2] via P2P/CUMEM cnode7-013:736484:745381 [3] NCCL INFO Channel 08/0 : 11[3] -> 10[2] via P2P/CUMEM cnode7-013:736484:745381 [3] NCCL INFO Channel 10/0 : 11[3] -> 10[2] via P2P/CUMEM cnode7-013:736484:745381 [3] NCCL INFO Channel 12/0 : 11[3] -> 10[2] via P2P/CUMEM cnode7-013:736484:745381 [3] NCCL INFO Channel 14/0 : 11[3] -> 10[2] via P2P/CUMEM cnode7-013:736487:746330 [6] NCCL INFO Connected all rings cnode7-013:736487:746330 [6] NCCL INFO Channel 01/0 : 14[6] -> 15[7] via P2P/CUMEM cnode7-013:736487:746330 [6] NCCL INFO Channel 03/0 : 14[6] -> 15[7] via P2P/CUMEM cnode7-013:736487:746330 [6] NCCL INFO Channel 05/0 : 14[6] -> 15[7] via P2P/CUMEM cnode7-013:736487:746330 [6] NCCL INFO Channel 06/0 : 14[6] -> 15[7] via P2P/CUMEM cnode7-013:736487:746330 [6] NCCL INFO Channel 09/0 : 14[6] -> 15[7] via P2P/CUMEM cnode7-013:736487:746330 [6] NCCL INFO Channel 11/0 : 14[6] -> 15[7] via P2P/CUMEM cnode7-013:736487:746330 [6] NCCL INFO Channel 13/0 : 14[6] -> 15[7] via P2P/CUMEM cnode7-013:736487:746330 [6] NCCL INFO Channel 14/0 : 14[6] -> 15[7] via P2P/CUMEM cnode7-013:736487:746330 [6] NCCL INFO Channel 06/0 : 6[6] -> 14[6] [receive] via NET/IB/7/GDRDMA cnode7-013:736487:746330 [6] NCCL INFO Channel 14/0 : 6[6] -> 14[6] [receive] via NET/IB/7/GDRDMA cnode7-013:736487:746330 [6] NCCL INFO Channel 00/0 : 14[6] -> 13[5] via P2P/CUMEM cnode7-013:736487:746330 [6] NCCL INFO Channel 02/0 : 14[6] -> 13[5] via P2P/CUMEM cnode7-013:736487:746330 [6] NCCL INFO Channel 04/0 : 14[6] -> 13[5] via P2P/CUMEM cnode7-013:736487:746330 [6] NCCL INFO Channel 08/0 : 14[6] -> 13[5] via P2P/CUMEM cnode7-013:736487:746330 [6] NCCL INFO Channel 10/0 : 14[6] -> 13[5] via P2P/CUMEM cnode7-013:736487:746330 [6] NCCL INFO Channel 12/0 : 14[6] -> 13[5] via P2P/CUMEM cnode7-013:736481:744977 [0] NCCL INFO Channel 05/0 : 8[0] -> 9[1] via P2P/CUMEM cnode7-013:736481:744977 [0] NCCL INFO Channel 07/0 : 8[0] -> 9[1] via P2P/CUMEM cnode7-013:736481:744977 [0] NCCL INFO Channel 08/0 : 8[0] -> 9[1] via P2P/CUMEM cnode7-013:736481:744977 [0] NCCL INFO Channel 11/0 : 8[0] -> 9[1] via P2P/CUMEM cnode7-013:736481:744977 [0] NCCL INFO Channel 13/0 : 8[0] -> 9[1] via P2P/CUMEM cnode7-013:736481:744977 [0] NCCL INFO Channel 15/0 : 8[0] -> 9[1] via P2P/CUMEM cnode7-013:736481:744977 [0] NCCL INFO Channel 02/0 : 8[0] -> 15[7] via P2P/CUMEM cnode7-013:736481:744977 [0] NCCL INFO Channel 04/0 : 8[0] -> 15[7] via P2P/CUMEM cnode7-013:736481:744977 [0] NCCL INFO Channel 06/0 : 8[0] -> 15[7] via P2P/CUMEM cnode7-013:736481:744977 [0] NCCL INFO Channel 10/0 : 8[0] -> 15[7] via P2P/CUMEM cnode7-013:736481:744977 [0] NCCL INFO Channel 12/0 : 8[0] -> 15[7] via P2P/CUMEM cnode7-013:736481:744977 [0] NCCL INFO Channel 14/0 : 8[0] -> 15[7] via P2P/CUMEM cnode7-013:736481:744977 [0] NCCL INFO Channel 00/0 : 0[0] -> 8[0] [receive] via NET/IB/0/GDRDMA cnode7-013:736481:744977 [0] NCCL INFO Channel 08/0 : 0[0] -> 8[0] [receive] via NET/IB/0/GDRDMA cnode7-013:736485:745370 [4] NCCL INFO Connected all rings cnode7-013:736485:745370 [4] NCCL INFO Channel 01/0 : 12[4] -> 13[5] via P2P/CUMEM cnode7-013:736485:745370 [4] NCCL INFO Channel 03/0 : 12[4] -> 13[5] via P2P/CUMEM cnode7-013:736485:745370 [4] NCCL INFO Channel 04/0 : 12[4] -> 13[5] via P2P/CUMEM cnode7-013:736485:745370 [4] NCCL INFO Channel 07/0 : 12[4] -> 13[5] via P2P/CUMEM cnode7-013:736485:745370 [4] NCCL INFO Channel 09/0 : 12[4] -> 13[5] via P2P/CUMEM cnode7-013:736485:745370 [4] NCCL INFO Channel 11/0 : 12[4] -> 13[5] via P2P/CUMEM cnode7-013:736485:745370 [4] NCCL INFO Channel 12/0 : 12[4] -> 13[5] via P2P/CUMEM cnode7-013:736485:745370 [4] NCCL INFO Channel 15/0 : 12[4] -> 13[5] via P2P/CUMEM cnode7-013:736485:745370 [4] NCCL INFO Channel 04/0 : 4[4] -> 12[4] [receive] via NET/IB/4/GDRDMA cnode7-013:736485:745370 [4] NCCL INFO Channel 12/0 : 4[4] -> 12[4] [receive] via NET/IB/4/GDRDMA cnode7-013:736485:745370 [4] NCCL INFO Channel 00/0 : 12[4] -> 11[3] via P2P/CUMEM cnode7-013:736485:745370 [4] NCCL INFO Channel 02/0 : 12[4] -> 11[3] via P2P/CUMEM cnode7-013:736485:745370 [4] NCCL INFO Channel 06/0 : 12[4] -> 11[3] via P2P/CUMEM cnode7-013:736485:745370 [4] NCCL INFO Channel 08/0 : 12[4] -> 11[3] via P2P/CUMEM cnode7-013:736485:745370 [4] NCCL INFO Channel 10/0 : 12[4] -> 11[3] via P2P/CUMEM cnode7-013:736485:745370 [4] NCCL INFO Channel 14/0 : 12[4] -> 11[3] via P2P/CUMEM cnode7-013:736486:745644 [5] NCCL INFO Connected all rings cnode7-013:736486:745644 [5] NCCL INFO Channel 01/0 : 13[5] -> 14[6] via P2P/CUMEM cnode7-013:736486:745644 [5] NCCL INFO Channel 03/0 : 13[5] -> 14[6] via P2P/CUMEM cnode7-013:736486:745644 [5] NCCL INFO Channel 05/0 : 13[5] -> 14[6] via P2P/CUMEM cnode7-013:736486:745644 [5] NCCL INFO Channel 07/0 : 13[5] -> 14[6] via P2P/CUMEM cnode7-013:736486:745644 [5] NCCL INFO Channel 09/0 : 13[5] -> 14[6] via P2P/CUMEM cnode7-013:736486:745644 [5] NCCL INFO Channel 11/0 : 13[5] -> 14[6] via P2P/CUMEM cnode7-013:736486:745644 [5] NCCL INFO Channel 13/0 : 13[5] -> 14[6] via P2P/CUMEM cnode7-013:736486:745644 [5] NCCL INFO Channel 15/0 : 13[5] -> 14[6] via P2P/CUMEM cnode7-013:736486:745644 [5] NCCL INFO Channel 05/0 : 5[5] -> 13[5] [receive] via NET/IB/6/GDRDMA cnode7-013:736486:745644 [5] NCCL INFO Channel 13/0 : 5[5] -> 13[5] [receive] via NET/IB/6/GDRDMA cnode7-013:736486:745644 [5] NCCL INFO Channel 00/0 : 13[5] -> 12[4] via P2P/CUMEM cnode7-013:736486:745644 [5] NCCL INFO Channel 02/0 : 13[5] -> 12[4] via P2P/CUMEM cnode7-013:736486:745644 [5] NCCL INFO Channel 04/0 : 13[5] -> 12[4] via P2P/CUMEM cnode7-013:736486:745644 [5] NCCL INFO Channel 06/0 : 13[5] -> 12[4] via P2P/CUMEM cnode7-013:736486:745644 [5] NCCL INFO Channel 08/0 : 13[5] -> 12[4] via P2P/CUMEM cnode7-013:736486:745644 [5] NCCL INFO Channel 10/0 : 13[5] -> 12[4] via P2P/CUMEM cnode7-013:736486:745644 [5] NCCL INFO Channel 12/0 : 13[5] -> 12[4] via P2P/CUMEM cnode7-013:736486:745644 [5] NCCL INFO Channel 14/0 : 13[5] -> 12[4] via P2P/CUMEM cnode7-012:1272012:1281554 [7] NCCL INFO Connected all rings cnode7-012:1272012:1281554 [7] NCCL INFO Channel 07/0 : 7[7] -> 15[7] [send] via NET/IB/8/GDRDMA cnode7-012:1272012:1281554 [7] NCCL INFO Channel 15/0 : 7[7] -> 15[7] [send] via NET/IB/8/GDRDMA cnode7-012:1272012:1281554 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/CUMEM cnode7-012:1272012:1281554 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/CUMEM cnode7-012:1272012:1281554 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/CUMEM cnode7-012:1272012:1281554 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/CUMEM cnode7-012:1272012:1281554 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/CUMEM cnode7-012:1272012:1281554 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/CUMEM cnode7-012:1272012:1281554 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/CUMEM cnode7-012:1272012:1281554 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/CUMEM cnode7-012:1272012:1281554 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/CUMEM cnode7-012:1272012:1281554 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/CUMEM cnode7-012:1272012:1281554 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/CUMEM cnode7-012:1272012:1281554 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/CUMEM cnode7-012:1272012:1281554 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/CUMEM cnode7-012:1272012:1281554 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/CUMEM cnode7-012:1272009:1281281 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/CUMEM cnode7-012:1272009:1281281 [4] NCCL INFO Channel 04/0 : 4[4] -> 12[4] [send] via NET/IB/4/GDRDMA cnode7-012:1272009:1281281 [4] NCCL INFO Channel 12/0 : 4[4] -> 12[4] [send] via NET/IB/4/GDRDMA cnode7-012:1272009:1281281 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/CUMEM cnode7-012:1272009:1281281 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/CUMEM cnode7-012:1272009:1281281 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/CUMEM cnode7-012:1272009:1281281 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/CUMEM cnode7-012:1272009:1281281 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/CUMEM cnode7-012:1272009:1281281 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/CUMEM cnode7-012:1272009:1281281 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/CUMEM cnode7-012:1272009:1281281 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/CUMEM cnode7-012:1272011:1281645 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/CUMEM cnode7-012:1272011:1281645 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/CUMEM cnode7-012:1272011:1281645 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/CUMEM cnode7-012:1272011:1281645 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/CUMEM cnode7-012:1272011:1281645 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/CUMEM cnode7-012:1272011:1281645 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/CUMEM cnode7-012:1272011:1281645 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/CUMEM cnode7-012:1272011:1281645 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/CUMEM cnode7-012:1272010:1281060 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/CUMEM cnode7-012:1272010:1281060 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/CUMEM cnode7-012:1272010:1281060 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/CUMEM cnode7-012:1272010:1281060 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/CUMEM cnode7-012:1272010:1281060 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/CUMEM cnode7-012:1272010:1281060 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/CUMEM cnode7-012:1272010:1281060 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/CUMEM cnode7-012:1272010:1281060 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/CUMEM cnode7-013:736488:746041 [7] NCCL INFO Channel 14/0 : 15[7] -> 14[6] via P2P/CUMEM cnode7-012:1272008:1281624 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/CUMEM cnode7-012:1272008:1281624 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/CUMEM cnode7-012:1272008:1281624 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/CUMEM cnode7-012:1272008:1281624 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/CUMEM cnode7-012:1272008:1281624 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/CUMEM cnode7-012:1272008:1281624 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/CUMEM cnode7-012:1272008:1281624 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/CUMEM cnode7-012:1272008:1281624 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/CUMEM cnode7-012:1272007:1281061 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/CUMEM cnode7-012:1272007:1281061 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/CUMEM cnode7-012:1272007:1281061 [2] NCCL INFO Connected all trees cnode7-013:736483:745475 [2] NCCL INFO Connected all trees cnode7-012:1272008:1281624 [3] NCCL INFO Connected all trees cnode7-013:736484:745381 [3] NCCL INFO Connected all trees cnode7-012:1272008:1281624 [3] NCCL INFO NVLS comm 0x5555e39b2180 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736 cnode7-012:1272007:1281061 [2] NCCL INFO NVLS comm 0x5555e39b9f60 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736 cnode7-013:736483:745475 [2] NCCL INFO NVLS comm 0x5555cf9ca470 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736 cnode7-013:736484:745381 [3] NCCL INFO NVLS comm 0x5555cd20c9e0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736 cnode7-012:1272009:1281281 [4] NCCL INFO Connected all trees cnode7-012:1272010:1281060 [5] NCCL INFO Connected all trees cnode7-012:1272011:1281645 [6] NCCL INFO Connected all trees cnode7-012:1272012:1281554 [7] NCCL INFO Connected all trees cnode7-013:736485:745370 [4] NCCL INFO Connected all trees cnode7-012:1272006:1281620 [1] NCCL INFO Connected all trees cnode7-013:736487:746330 [6] NCCL INFO Connected all trees cnode7-013:736488:746041 [7] NCCL INFO Connected all trees cnode7-013:736486:745644 [5] NCCL INFO Connected all trees cnode7-013:736481:744977 [0] NCCL INFO Connected all trees cnode7-013:736482:746397 [1] NCCL INFO Connected all trees cnode7-012:1272005:1281059 [0] NCCL INFO Connected all trees cnode7-012:1272010:1281060 [5] NCCL INFO NVLS comm 0x5555e39cd100 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736 cnode7-012:1272005:1281059 [0] NCCL INFO NVLS comm 0x5555c2f5fad0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736 cnode7-013:736488:746041 [7] NCCL INFO NVLS comm 0x5555cf9aad40 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736 cnode7-013:736482:746397 [1] NCCL INFO NVLS comm 0x5555cf9ca2c0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736 cnode7-013:736487:746330 [6] NCCL INFO NVLS comm 0x5555cf9ae150 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736 cnode7-013:736481:744977 [0] NCCL INFO NVLS comm 0x5555c59f0f70 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736 cnode7-013:736485:745370 [4] NCCL INFO NVLS comm 0x5555cf9ca4a0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736 cnode7-013:736486:745644 [5] NCCL INFO NVLS comm 0x5555cd22bfa0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736 cnode7-012:1272012:1281554 [7] NCCL INFO NVLS comm 0x5555cf991e20 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736 cnode7-012:1272009:1281281 [4] NCCL INFO NVLS comm 0x5555cd223a50 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736 cnode7-012:1272006:1281620 [1] NCCL INFO NVLS comm 0x5555cf9b5e30 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736 cnode7-012:1272011:1281645 [6] NCCL INFO NVLS comm 0x5555cf994130 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736 cnode7-013:736483:745475 [2] NCCL INFO Channel 00/0 : 2[2] -> 10[2] [receive] via NET/IB/2/GDRDMA cnode7-013:736483:745475 [2] NCCL INFO Channel 01/0 : 2[2] -> 10[2] [receive] via NET/IB/2/GDRDMA cnode7-013:736483:745475 [2] NCCL INFO Channel 04/0 : 2[2] -> 10[2] [receive] via NET/IB/2/GDRDMA cnode7-013:736483:745475 [2] NCCL INFO Channel 05/0 : 2[2] -> 10[2] [receive] via NET/IB/2/GDRDMA cnode7-013:736483:745475 [2] NCCL INFO Channel 06/0 : 2[2] -> 10[2] [receive] via NET/IB/2/GDRDMA cnode7-013:736483:745475 [2] NCCL INFO Channel 07/0 : 2[2] -> 10[2] [receive] via NET/IB/2/GDRDMA cnode7-013:736483:745475 [2] NCCL INFO Channel 08/0 : 2[2] -> 10[2] [receive] via NET/IB/2/GDRDMA cnode7-013:736483:745475 [2] NCCL INFO Channel 09/0 : 2[2] -> 10[2] [receive] via NET/IB/2/GDRDMA cnode7-013:736483:745475 [2] NCCL INFO Channel 12/0 : 2[2] -> 10[2] [receive] via NET/IB/2/GDRDMA cnode7-013:736483:745475 [2] NCCL INFO Channel 13/0 : 2[2] -> 10[2] [receive] via NET/IB/2/GDRDMA cnode7-013:736483:745475 [2] NCCL INFO Channel 14/0 : 2[2] -> 10[2] [receive] via NET/IB/2/GDRDMA cnode7-013:736483:745475 [2] NCCL INFO Channel 15/0 : 2[2] -> 10[2] [receive] via NET/IB/2/GDRDMA cnode7-013:736483:745475 [2] NCCL INFO Channel 00/0 : 10[2] -> 2[2] [send] via NET/IB/2/GDRDMA cnode7-013:736483:745475 [2] NCCL INFO Channel 01/0 : 10[2] -> 2[2] [send] via NET/IB/2/GDRDMA cnode7-012:1272010:1281060 [5] NCCL INFO Channel 00/0 : 13[5] -> 5[5] [receive] via NET/IB/6/GDRDMA cnode7-012:1272010:1281060 [5] NCCL INFO Channel 01/0 : 13[5] -> 5[5] [receive] via NET/IB/6/GDRDMA cnode7-012:1272010:1281060 [5] NCCL INFO Channel 02/0 : 13[5] -> 5[5] [receive] via NET/IB/6/GDRDMA cnode7-012:1272010:1281060 [5] NCCL INFO Channel 03/0 : 13[5] -> 5[5] [receive] via NET/IB/6/GDRDMA cnode7-012:1272010:1281060 [5] NCCL INFO Channel 04/0 : 13[5] -> 5[5] [receive] via NET/IB/6/GDRDMA cnode7-012:1272010:1281060 [5] NCCL INFO Channel 06/0 : 13[5] -> 5[5] [receive] via NET/IB/6/GDRDMA cnode7-012:1272010:1281060 [5] NCCL INFO Channel 07/0 : 13[5] -> 5[5] [receive] via NET/IB/6/GDRDMA cnode7-012:1272010:1281060 [5] NCCL INFO Channel 08/0 : 13[5] -> 5[5] [receive] via NET/IB/6/GDRDMA cnode7-012:1272010:1281060 [5] NCCL INFO Channel 09/0 : 13[5] -> 5[5] [receive] via NET/IB/6/GDRDMA cnode7-012:1272010:1281060 [5] NCCL INFO Channel 10/0 : 13[5] -> 5[5] [receive] via NET/IB/6/GDRDMA cnode7-012:1272010:1281060 [5] NCCL INFO Channel 11/0 : 13[5] -> 5[5] [receive] via NET/IB/6/GDRDMA cnode7-012:1272010:1281060 [5] NCCL INFO Channel 12/0 : 13[5] -> 5[5] [receive] via NET/IB/6/GDRDMA cnode7-012:1272010:1281060 [5] NCCL INFO Channel 14/0 : 13[5] -> 5[5] [receive] via NET/IB/6/GDRDMA cnode7-012:1272010:1281060 [5] NCCL INFO Channel 15/0 : 13[5] -> 5[5] [receive] via NET/IB/6/GDRDMA cnode7-012:1272010:1281060 [5] NCCL INFO Channel 00/0 : 5[5] -> 13[5] [send] via NET/IB/6/GDRDMA cnode7-012:1272010:1281060 [5] NCCL INFO Channel 01/0 : 5[5] -> 13[5] [send] via NET/IB/6/GDRDMA cnode7-012:1272010:1281060 [5] NCCL INFO Channel 02/0 : 5[5] -> 13[5] [send] via NET/IB/6/GDRDMA cnode7-012:1272010:1281060 [5] NCCL INFO Channel 03/0 : 5[5] -> 13[5] [send] via NET/IB/6/GDRDMA cnode7-012:1272010:1281060 [5] NCCL INFO Channel 06/0 : 5[5] -> 13[5] [send] via NET/IB/6/GDRDMA cnode7-012:1272010:1281060 [5] NCCL INFO Channel 07/0 : 5[5] -> 13[5] [send] via NET/IB/6/GDRDMA cnode7-012:1272010:1281060 [5] NCCL INFO Channel 08/0 : 5[5] -> 13[5] [send] via NET/IB/6/GDRDMA cnode7-012:1272010:1281060 [5] NCCL INFO Channel 09/0 : 5[5] -> 13[5] [send] via NET/IB/6/GDRDMA cnode7-012:1272010:1281060 [5] NCCL INFO Channel 10/0 : 5[5] -> 13[5] [send] via NET/IB/6/GDRDMA cnode7-012:1272010:1281060 [5] NCCL INFO Channel 11/0 : 5[5] -> 13[5] [send] via NET/IB/6/GDRDMA cnode7-012:1272010:1281060 [5] NCCL INFO Channel 14/0 : 5[5] -> 13[5] [send] via NET/IB/6/GDRDMA cnode7-012:1272005:1281059 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [receive] via NET/IB/0/GDRDMA cnode7-012:1272005:1281059 [0] NCCL INFO Channel 02/0 : 8[0] -> 0[0] [receive] via NET/IB/0/GDRDMA cnode7-012:1272005:1281059 [0] NCCL INFO Channel 03/0 : 8[0] -> 0[0] [receive] via NET/IB/0/GDRDMA cnode7-012:1272005:1281059 [0] NCCL INFO Channel 04/0 : 8[0] -> 0[0] [receive] via NET/IB/0/GDRDMA cnode7-012:1272005:1281059 [0] NCCL INFO Channel 05/0 : 8[0] -> 0[0] [receive] via NET/IB/0/GDRDMA cnode7-012:1272005:1281059 [0] NCCL INFO Channel 06/0 : 8[0] -> 0[0] [receive] via NET/IB/0/GDRDMA cnode7-012:1272005:1281059 [0] NCCL INFO Channel 07/0 : 8[0] -> 0[0] [receive] via NET/IB/0/GDRDMA cnode7-012:1272005:1281059 [0] NCCL INFO Channel 09/0 : 8[0] -> 0[0] [receive] via NET/IB/0/GDRDMA cnode7-012:1272005:1281059 [0] NCCL INFO Channel 10/0 : 8[0] -> 0[0] [receive] via NET/IB/0/GDRDMA cnode7-012:1272005:1281059 [0] NCCL INFO Channel 11/0 : 8[0] -> 0[0] [receive] via NET/IB/0/GDRDMA cnode7-012:1272005:1281059 [0] NCCL INFO Channel 12/0 : 8[0] -> 0[0] [receive] via NET/IB/0/GDRDMA cnode7-012:1272005:1281059 [0] NCCL INFO Channel 13/0 : 8[0] -> 0[0] [receive] via NET/IB/0/GDRDMA cnode7-012:1272005:1281059 [0] NCCL INFO Channel 14/0 : 8[0] -> 0[0] [receive] via NET/IB/0/GDRDMA cnode7-012:1272005:1281059 [0] NCCL INFO Channel 15/0 : 8[0] -> 0[0] [receive] via NET/IB/0/GDRDMA cnode7-012:1272005:1281059 [0] NCCL INFO Channel 02/0 : 0[0] -> 8[0] [send] via NET/IB/0/GDRDMA cnode7-012:1272005:1281059 [0] NCCL INFO Channel 03/0 : 0[0] -> 8[0] [send] via NET/IB/0/GDRDMA cnode7-012:1272005:1281059 [0] NCCL INFO Channel 04/0 : 0[0] -> 8[0] [send] via NET/IB/0/GDRDMA cnode7-012:1272005:1281059 [0] NCCL INFO Channel 05/0 : 0[0] -> 8[0] [send] via NET/IB/0/GDRDMA cnode7-012:1272005:1281059 [0] NCCL INFO Channel 06/0 : 0[0] -> 8[0] [send] via NET/IB/0/GDRDMA cnode7-012:1272005:1281059 [0] NCCL INFO Channel 07/0 : 0[0] -> 8[0] [send] via NET/IB/0/GDRDMA cnode7-012:1272005:1281059 [0] NCCL INFO Channel 10/0 : 0[0] -> 8[0] [send] via NET/IB/0/GDRDMA cnode7-012:1272005:1281059 [0] NCCL INFO Channel 11/0 : 0[0] -> 8[0] [send] via NET/IB/0/GDRDMA cnode7-012:1272005:1281059 [0] NCCL INFO Channel 12/0 : 0[0] -> 8[0] [send] via NET/IB/0/GDRDMA cnode7-012:1272005:1281059 [0] NCCL INFO Channel 13/0 : 0[0] -> 8[0] [send] via NET/IB/0/GDRDMA cnode7-012:1272005:1281059 [0] NCCL INFO Channel 14/0 : 0[0] -> 8[0] [send] via NET/IB/0/GDRDMA cnode7-012:1272005:1281059 [0] NCCL INFO Channel 15/0 : 0[0] -> 8[0] [send] via NET/IB/0/GDRDMA cnode7-013:736488:746041 [7] NCCL INFO Channel 00/0 : 7[7] -> 15[7] [receive] via NET/IB/8/GDRDMA cnode7-013:736488:746041 [7] NCCL INFO Channel 01/0 : 7[7] -> 15[7] [receive] via NET/IB/8/GDRDMA cnode7-013:736488:746041 [7] NCCL INFO Channel 02/0 : 7[7] -> 15[7] [receive] via NET/IB/8/GDRDMA cnode7-013:736488:746041 [7] NCCL INFO Channel 03/0 : 7[7] -> 15[7] [receive] via NET/IB/8/GDRDMA cnode7-013:736488:746041 [7] NCCL INFO Channel 04/0 : 7[7] -> 15[7] [receive] via NET/IB/8/GDRDMA cnode7-013:736488:746041 [7] NCCL INFO Channel 05/0 : 7[7] -> 15[7] [receive] via NET/IB/8/GDRDMA cnode7-013:736488:746041 [7] NCCL INFO Channel 08/0 : 7[7] -> 15[7] [receive] via NET/IB/8/GDRDMA cnode7-013:736488:746041 [7] NCCL INFO Channel 09/0 : 7[7] -> 15[7] [receive] via NET/IB/8/GDRDMA cnode7-013:736488:746041 [7] NCCL INFO Channel 10/0 : 7[7] -> 15[7] [receive] via NET/IB/8/GDRDMA cnode7-013:736488:746041 [7] NCCL INFO Channel 11/0 : 7[7] -> 15[7] [receive] via NET/IB/8/GDRDMA cnode7-012:1272008:1281624 [3] NCCL INFO Channel 00/0 : 11[3] -> 3[3] [receive] via NET/IB/3/GDRDMA cnode7-012:1272008:1281624 [3] NCCL INFO Channel 01/0 : 11[3] -> 3[3] [receive] via NET/IB/3/GDRDMA cnode7-012:1272008:1281624 [3] NCCL INFO Channel 02/0 : 11[3] -> 3[3] [receive] via NET/IB/3/GDRDMA cnode7-012:1272008:1281624 [3] NCCL INFO Channel 04/0 : 11[3] -> 3[3] [receive] via NET/IB/3/GDRDMA cnode7-012:1272008:1281624 [3] NCCL INFO Channel 05/0 : 11[3] -> 3[3] [receive] via NET/IB/3/GDRDMA cnode7-012:1272008:1281624 [3] NCCL INFO Channel 06/0 : 11[3] -> 3[3] [receive] via NET/IB/3/GDRDMA cnode7-012:1272008:1281624 [3] NCCL INFO Channel 07/0 : 11[3] -> 3[3] [receive] via NET/IB/3/GDRDMA cnode7-012:1272008:1281624 [3] NCCL INFO Channel 08/0 : 11[3] -> 3[3] [receive] via NET/IB/3/GDRDMA cnode7-012:1272008:1281624 [3] NCCL INFO Channel 09/0 : 11[3] -> 3[3] [receive] via NET/IB/3/GDRDMA cnode7-012:1272008:1281624 [3] NCCL INFO Channel 10/0 : 11[3] -> 3[3] [receive] via NET/IB/3/GDRDMA cnode7-012:1272008:1281624 [3] NCCL INFO Channel 12/0 : 11[3] -> 3[3] [receive] via NET/IB/3/GDRDMA cnode7-012:1272008:1281624 [3] NCCL INFO Channel 13/0 : 11[3] -> 3[3] [receive] via NET/IB/3/GDRDMA cnode7-012:1272008:1281624 [3] NCCL INFO Channel 14/0 : 11[3] -> 3[3] [receive] via NET/IB/3/GDRDMA cnode7-012:1272008:1281624 [3] NCCL INFO Channel 15/0 : 11[3] -> 3[3] [receive] via NET/IB/3/GDRDMA cnode7-012:1272008:1281624 [3] NCCL INFO Channel 00/0 : 3[3] -> 11[3] [send] via NET/IB/3/GDRDMA cnode7-012:1272008:1281624 [3] NCCL INFO Channel 01/0 : 3[3] -> 11[3] [send] via NET/IB/3/GDRDMA cnode7-012:1272008:1281624 [3] NCCL INFO Channel 04/0 : 3[3] -> 11[3] [send] via NET/IB/3/GDRDMA cnode7-012:1272008:1281624 [3] NCCL INFO Channel 05/0 : 3[3] -> 11[3] [send] via NET/IB/3/GDRDMA cnode7-012:1272008:1281624 [3] NCCL INFO Channel 06/0 : 3[3] -> 11[3] [send] via NET/IB/3/GDRDMA cnode7-012:1272008:1281624 [3] NCCL INFO Channel 07/0 : 3[3] -> 11[3] [send] via NET/IB/3/GDRDMA cnode7-012:1272008:1281624 [3] NCCL INFO Channel 08/0 : 3[3] -> 11[3] [send] via NET/IB/3/GDRDMA cnode7-012:1272008:1281624 [3] NCCL INFO Channel 09/0 : 3[3] -> 11[3] [send] via NET/IB/3/GDRDMA cnode7-012:1272008:1281624 [3] NCCL INFO Channel 12/0 : 3[3] -> 11[3] [send] via NET/IB/3/GDRDMA cnode7-012:1272008:1281624 [3] NCCL INFO Channel 13/0 : 3[3] -> 11[3] [send] via NET/IB/3/GDRDMA cnode7-012:1272008:1281624 [3] NCCL INFO Channel 14/0 : 3[3] -> 11[3] [send] via NET/IB/3/GDRDMA cnode7-012:1272008:1281624 [3] NCCL INFO Channel 15/0 : 3[3] -> 11[3] [send] via NET/IB/3/GDRDMA cnode7-013:736488:746041 [7] NCCL INFO Channel 12/0 : 7[7] -> 15[7] [receive] via NET/IB/8/GDRDMA cnode7-013:736488:746041 [7] NCCL INFO Channel 13/0 : 7[7] -> 15[7] [receive] via NET/IB/8/GDRDMA cnode7-013:736488:746041 [7] NCCL INFO Channel 00/0 : 15[7] -> 7[7] [send] via NET/IB/8/GDRDMA cnode7-013:736488:746041 [7] NCCL INFO Channel 01/0 : 15[7] -> 7[7] [send] via NET/IB/8/GDRDMA cnode7-013:736488:746041 [7] NCCL INFO Channel 02/0 : 15[7] -> 7[7] [send] via NET/IB/8/GDRDMA cnode7-013:736488:746041 [7] NCCL INFO Channel 03/0 : 15[7] -> 7[7] [send] via NET/IB/8/GDRDMA cnode7-013:736488:746041 [7] NCCL INFO Channel 04/0 : 15[7] -> 7[7] [send] via NET/IB/8/GDRDMA cnode7-013:736488:746041 [7] NCCL INFO Channel 05/0 : 15[7] -> 7[7] [send] via NET/IB/8/GDRDMA cnode7-013:736488:746041 [7] NCCL INFO Channel 06/0 : 15[7] -> 7[7] [send] via NET/IB/8/GDRDMA cnode7-013:736488:746041 [7] NCCL INFO Channel 08/0 : 15[7] -> 7[7] [send] via NET/IB/8/GDRDMA cnode7-013:736488:746041 [7] NCCL INFO Channel 09/0 : 15[7] -> 7[7] [send] via NET/IB/8/GDRDMA cnode7-013:736488:746041 [7] NCCL INFO Channel 10/0 : 15[7] -> 7[7] [send] via NET/IB/8/GDRDMA cnode7-013:736488:746041 [7] NCCL INFO Channel 11/0 : 15[7] -> 7[7] [send] via NET/IB/8/GDRDMA cnode7-013:736488:746041 [7] NCCL INFO Channel 12/0 : 15[7] -> 7[7] [send] via NET/IB/8/GDRDMA cnode7-013:736488:746041 [7] NCCL INFO Channel 13/0 : 15[7] -> 7[7] [send] via NET/IB/8/GDRDMA cnode7-013:736488:746041 [7] NCCL INFO Channel 14/0 : 15[7] -> 7[7] [send] via NET/IB/8/GDRDMA cnode7-013:736482:746397 [1] NCCL INFO Channel 02/0 : 1[1] -> 9[1] [receive] via NET/IB/1/GDRDMA cnode7-013:736482:746397 [1] NCCL INFO Channel 03/0 : 1[1] -> 9[1] [receive] via NET/IB/1/GDRDMA cnode7-013:736482:746397 [1] NCCL INFO Channel 04/0 : 1[1] -> 9[1] [receive] via NET/IB/1/GDRDMA cnode7-013:736482:746397 [1] NCCL INFO Channel 05/0 : 1[1] -> 9[1] [receive] via NET/IB/1/GDRDMA cnode7-013:736482:746397 [1] NCCL INFO Channel 06/0 : 1[1] -> 9[1] [receive] via NET/IB/1/GDRDMA cnode7-013:736482:746397 [1] NCCL INFO Channel 07/0 : 1[1] -> 9[1] [receive] via NET/IB/1/GDRDMA cnode7-013:736482:746397 [1] NCCL INFO Channel 10/0 : 1[1] -> 9[1] [receive] via NET/IB/1/GDRDMA cnode7-013:736482:746397 [1] NCCL INFO Channel 11/0 : 1[1] -> 9[1] [receive] via NET/IB/1/GDRDMA cnode7-013:736482:746397 [1] NCCL INFO Channel 12/0 : 1[1] -> 9[1] [receive] via NET/IB/1/GDRDMA cnode7-013:736482:746397 [1] NCCL INFO Channel 13/0 : 1[1] -> 9[1] [receive] via NET/IB/1/GDRDMA cnode7-013:736482:746397 [1] NCCL INFO Channel 14/0 : 1[1] -> 9[1] [receive] via NET/IB/1/GDRDMA cnode7-013:736482:746397 [1] NCCL INFO Channel 15/0 : 1[1] -> 9[1] [receive] via NET/IB/1/GDRDMA cnode7-013:736482:746397 [1] NCCL INFO Channel 00/0 : 9[1] -> 1[1] [send] via NET/IB/1/GDRDMA cnode7-013:736482:746397 [1] NCCL INFO Channel 02/0 : 9[1] -> 1[1] [send] via NET/IB/1/GDRDMA cnode7-013:736482:746397 [1] NCCL INFO Channel 03/0 : 9[1] -> 1[1] [send] via NET/IB/1/GDRDMA cnode7-013:736482:746397 [1] NCCL INFO Channel 04/0 : 9[1] -> 1[1] [send] via NET/IB/1/GDRDMA cnode7-013:736482:746397 [1] NCCL INFO Channel 05/0 : 9[1] -> 1[1] [send] via NET/IB/1/GDRDMA cnode7-013:736482:746397 [1] NCCL INFO Channel 06/0 : 9[1] -> 1[1] [send] via NET/IB/1/GDRDMA cnode7-013:736482:746397 [1] NCCL INFO Channel 07/0 : 9[1] -> 1[1] [send] via NET/IB/1/GDRDMA cnode7-013:736482:746397 [1] NCCL INFO Channel 08/0 : 9[1] -> 1[1] [send] via NET/IB/1/GDRDMA cnode7-013:736482:746397 [1] NCCL INFO Channel 10/0 : 9[1] -> 1[1] [send] via NET/IB/1/GDRDMA cnode7-013:736482:746397 [1] NCCL INFO Channel 11/0 : 9[1] -> 1[1] [send] via NET/IB/1/GDRDMA cnode7-013:736482:746397 [1] NCCL INFO Channel 12/0 : 9[1] -> 1[1] [send] via NET/IB/1/GDRDMA cnode7-013:736482:746397 [1] NCCL INFO Channel 13/0 : 9[1] -> 1[1] [send] via NET/IB/1/GDRDMA cnode7-013:736482:746397 [1] NCCL INFO Channel 14/0 : 9[1] -> 1[1] [send] via NET/IB/1/GDRDMA cnode7-013:736482:746397 [1] NCCL INFO Channel 15/0 : 9[1] -> 1[1] [send] via NET/IB/1/GDRDMA cnode7-012:1272007:1281061 [2] NCCL INFO Channel 00/0 : 10[2] -> 2[2] [receive] via NET/IB/2/GDRDMA cnode7-012:1272007:1281061 [2] NCCL INFO Channel 01/0 : 10[2] -> 2[2] [receive] via NET/IB/2/GDRDMA cnode7-012:1272007:1281061 [2] NCCL INFO Channel 03/0 : 10[2] -> 2[2] [receive] via NET/IB/2/GDRDMA cnode7-012:1272007:1281061 [2] NCCL INFO Channel 04/0 : 10[2] -> 2[2] [receive] via NET/IB/2/GDRDMA cnode7-012:1272007:1281061 [2] NCCL INFO Channel 05/0 : 10[2] -> 2[2] [receive] via NET/IB/2/GDRDMA cnode7-012:1272007:1281061 [2] NCCL INFO Channel 06/0 : 10[2] -> 2[2] [receive] via NET/IB/2/GDRDMA cnode7-012:1272007:1281061 [2] NCCL INFO Channel 07/0 : 10[2] -> 2[2] [receive] via NET/IB/2/GDRDMA cnode7-012:1272007:1281061 [2] NCCL INFO Channel 08/0 : 10[2] -> 2[2] [receive] via NET/IB/2/GDRDMA cnode7-012:1272007:1281061 [2] NCCL INFO Channel 09/0 : 10[2] -> 2[2] [receive] via NET/IB/2/GDRDMA cnode7-012:1272007:1281061 [2] NCCL INFO Channel 11/0 : 10[2] -> 2[2] [receive] via NET/IB/2/GDRDMA cnode7-012:1272007:1281061 [2] NCCL INFO Channel 12/0 : 10[2] -> 2[2] [receive] via NET/IB/2/GDRDMA cnode7-012:1272007:1281061 [2] NCCL INFO Channel 13/0 : 10[2] -> 2[2] [receive] via NET/IB/2/GDRDMA cnode7-012:1272007:1281061 [2] NCCL INFO Channel 14/0 : 10[2] -> 2[2] [receive] via NET/IB/2/GDRDMA cnode7-012:1272007:1281061 [2] NCCL INFO Channel 15/0 : 10[2] -> 2[2] [receive] via NET/IB/2/GDRDMA cnode7-012:1272007:1281061 [2] NCCL INFO Channel 00/0 : 2[2] -> 10[2] [send] via NET/IB/2/GDRDMA cnode7-012:1272007:1281061 [2] NCCL INFO Channel 01/0 : 2[2] -> 10[2] [send] via NET/IB/2/GDRDMA cnode7-012:1272007:1281061 [2] NCCL INFO Channel 04/0 : 2[2] -> 10[2] [send] via NET/IB/2/GDRDMA cnode7-012:1272007:1281061 [2] NCCL INFO Channel 05/0 : 2[2] -> 10[2] [send] via NET/IB/2/GDRDMA cnode7-012:1272007:1281061 [2] NCCL INFO Channel 06/0 : 2[2] -> 10[2] [send] via NET/IB/2/GDRDMA cnode7-012:1272007:1281061 [2] NCCL INFO Channel 07/0 : 2[2] -> 10[2] [send] via NET/IB/2/GDRDMA cnode7-012:1272007:1281061 [2] NCCL INFO Channel 08/0 : 2[2] -> 10[2] [send] via NET/IB/2/GDRDMA cnode7-012:1272007:1281061 [2] NCCL INFO Channel 09/0 : 2[2] -> 10[2] [send] via NET/IB/2/GDRDMA cnode7-012:1272007:1281061 [2] NCCL INFO Channel 12/0 : 2[2] -> 10[2] [send] via NET/IB/2/GDRDMA cnode7-012:1272007:1281061 [2] NCCL INFO Channel 13/0 : 2[2] -> 10[2] [send] via NET/IB/2/GDRDMA cnode7-012:1272007:1281061 [2] NCCL INFO Channel 14/0 : 2[2] -> 10[2] [send] via NET/IB/2/GDRDMA cnode7-012:1272007:1281061 [2] NCCL INFO Channel 15/0 : 2[2] -> 10[2] [send] via NET/IB/2/GDRDMA cnode7-013:736484:745381 [3] NCCL INFO Channel 00/0 : 3[3] -> 11[3] [receive] via NET/IB/3/GDRDMA cnode7-013:736484:745381 [3] NCCL INFO Channel 01/0 : 3[3] -> 11[3] [receive] via NET/IB/3/GDRDMA cnode7-013:736484:745381 [3] NCCL INFO Channel 04/0 : 3[3] -> 11[3] [receive] via NET/IB/3/GDRDMA cnode7-013:736484:745381 [3] NCCL INFO Channel 05/0 : 3[3] -> 11[3] [receive] via NET/IB/3/GDRDMA cnode7-013:736484:745381 [3] NCCL INFO Channel 06/0 : 3[3] -> 11[3] [receive] via NET/IB/3/GDRDMA cnode7-013:736484:745381 [3] NCCL INFO Channel 07/0 : 3[3] -> 11[3] [receive] via NET/IB/3/GDRDMA cnode7-013:736484:745381 [3] NCCL INFO Channel 08/0 : 3[3] -> 11[3] [receive] via NET/IB/3/GDRDMA cnode7-013:736484:745381 [3] NCCL INFO Channel 09/0 : 3[3] -> 11[3] [receive] via NET/IB/3/GDRDMA cnode7-013:736484:745381 [3] NCCL INFO Channel 12/0 : 3[3] -> 11[3] [receive] via NET/IB/3/GDRDMA cnode7-013:736484:745381 [3] NCCL INFO Channel 13/0 : 3[3] -> 11[3] [receive] via NET/IB/3/GDRDMA cnode7-013:736484:745381 [3] NCCL INFO Channel 14/0 : 3[3] -> 11[3] [receive] via NET/IB/3/GDRDMA cnode7-013:736484:745381 [3] NCCL INFO Channel 15/0 : 3[3] -> 11[3] [receive] via NET/IB/3/GDRDMA cnode7-013:736484:745381 [3] NCCL INFO Channel 00/0 : 11[3] -> 3[3] [send] via NET/IB/3/GDRDMA cnode7-013:736484:745381 [3] NCCL INFO Channel 01/0 : 11[3] -> 3[3] [send] via NET/IB/3/GDRDMA cnode7-013:736484:745381 [3] NCCL INFO Channel 02/0 : 11[3] -> 3[3] [send] via NET/IB/3/GDRDMA cnode7-013:736484:745381 [3] NCCL INFO Channel 04/0 : 11[3] -> 3[3] [send] via NET/IB/3/GDRDMA cnode7-013:736484:745381 [3] NCCL INFO Channel 05/0 : 11[3] -> 3[3] [send] via NET/IB/3/GDRDMA cnode7-013:736484:745381 [3] NCCL INFO Channel 06/0 : 11[3] -> 3[3] [send] via NET/IB/3/GDRDMA cnode7-013:736484:745381 [3] NCCL INFO Channel 07/0 : 11[3] -> 3[3] [send] via NET/IB/3/GDRDMA cnode7-013:736484:745381 [3] NCCL INFO Channel 08/0 : 11[3] -> 3[3] [send] via NET/IB/3/GDRDMA cnode7-013:736484:745381 [3] NCCL INFO Channel 09/0 : 11[3] -> 3[3] [send] via NET/IB/3/GDRDMA cnode7-013:736484:745381 [3] NCCL INFO Channel 10/0 : 11[3] -> 3[3] [send] via NET/IB/3/GDRDMA cnode7-013:736484:745381 [3] NCCL INFO Channel 12/0 : 11[3] -> 3[3] [send] via NET/IB/3/GDRDMA cnode7-013:736484:745381 [3] NCCL INFO Channel 13/0 : 11[3] -> 3[3] [send] via NET/IB/3/GDRDMA cnode7-013:736484:745381 [3] NCCL INFO Channel 14/0 : 11[3] -> 3[3] [send] via NET/IB/3/GDRDMA cnode7-013:736484:745381 [3] NCCL INFO Channel 15/0 : 11[3] -> 3[3] [send] via NET/IB/3/GDRDMA cnode7-013:736487:746330 [6] NCCL INFO Channel 00/0 : 6[6] -> 14[6] [receive] via NET/IB/7/GDRDMA cnode7-013:736487:746330 [6] NCCL INFO Channel 01/0 : 6[6] -> 14[6] [receive] via NET/IB/7/GDRDMA cnode7-013:736487:746330 [6] NCCL INFO Channel 02/0 : 6[6] -> 14[6] [receive] via NET/IB/7/GDRDMA cnode7-013:736487:746330 [6] NCCL INFO Channel 03/0 : 6[6] -> 14[6] [receive] via NET/IB/7/GDRDMA cnode7-013:736487:746330 [6] NCCL INFO Channel 04/0 : 6[6] -> 14[6] [receive] via NET/IB/7/GDRDMA cnode7-013:736487:746330 [6] NCCL INFO Channel 05/0 : 6[6] -> 14[6] [receive] via NET/IB/7/GDRDMA cnode7-013:736487:746330 [6] NCCL INFO Channel 08/0 : 6[6] -> 14[6] [receive] via NET/IB/7/GDRDMA cnode7-013:736487:746330 [6] NCCL INFO Channel 09/0 : 6[6] -> 14[6] [receive] via NET/IB/7/GDRDMA cnode7-013:736487:746330 [6] NCCL INFO Channel 10/0 : 6[6] -> 14[6] [receive] via NET/IB/7/GDRDMA cnode7-013:736487:746330 [6] NCCL INFO Channel 11/0 : 6[6] -> 14[6] [receive] via NET/IB/7/GDRDMA cnode7-013:736487:746330 [6] NCCL INFO Channel 12/0 : 6[6] -> 14[6] [receive] via NET/IB/7/GDRDMA cnode7-013:736487:746330 [6] NCCL INFO Channel 13/0 : 6[6] -> 14[6] [receive] via NET/IB/7/GDRDMA cnode7-013:736487:746330 [6] NCCL INFO Channel 00/0 : 14[6] -> 6[6] [send] via NET/IB/7/GDRDMA cnode7-013:736487:746330 [6] NCCL INFO Channel 01/0 : 14[6] -> 6[6] [send] via NET/IB/7/GDRDMA cnode7-013:736487:746330 [6] NCCL INFO Channel 02/0 : 14[6] -> 6[6] [send] via NET/IB/7/GDRDMA cnode7-013:736487:746330 [6] NCCL INFO Channel 03/0 : 14[6] -> 6[6] [send] via NET/IB/7/GDRDMA cnode7-013:736487:746330 [6] NCCL INFO Channel 04/0 : 14[6] -> 6[6] [send] via NET/IB/7/GDRDMA cnode7-013:736487:746330 [6] NCCL INFO Channel 05/0 : 14[6] -> 6[6] [send] via NET/IB/7/GDRDMA cnode7-013:736487:746330 [6] NCCL INFO Channel 07/0 : 14[6] -> 6[6] [send] via NET/IB/7/GDRDMA cnode7-013:736487:746330 [6] NCCL INFO Channel 08/0 : 14[6] -> 6[6] [send] via NET/IB/7/GDRDMA cnode7-013:736487:746330 [6] NCCL INFO Channel 09/0 : 14[6] -> 6[6] [send] via NET/IB/7/GDRDMA cnode7-013:736487:746330 [6] NCCL INFO Channel 10/0 : 14[6] -> 6[6] [send] via NET/IB/7/GDRDMA cnode7-013:736487:746330 [6] NCCL INFO Channel 11/0 : 14[6] -> 6[6] [send] via NET/IB/7/GDRDMA cnode7-013:736487:746330 [6] NCCL INFO Channel 12/0 : 14[6] -> 6[6] [send] via NET/IB/7/GDRDMA cnode7-013:736487:746330 [6] NCCL INFO Channel 13/0 : 14[6] -> 6[6] [send] via NET/IB/7/GDRDMA cnode7-013:736487:746330 [6] NCCL INFO Channel 15/0 : 14[6] -> 6[6] [send] via NET/IB/7/GDRDMA cnode7-013:736481:744977 [0] NCCL INFO Channel 02/0 : 0[0] -> 8[0] [receive] via NET/IB/0/GDRDMA cnode7-013:736481:744977 [0] NCCL INFO Channel 03/0 : 0[0] -> 8[0] [receive] via NET/IB/0/GDRDMA cnode7-013:736481:744977 [0] NCCL INFO Channel 04/0 : 0[0] -> 8[0] [receive] via NET/IB/0/GDRDMA cnode7-013:736481:744977 [0] NCCL INFO Channel 05/0 : 0[0] -> 8[0] [receive] via NET/IB/0/GDRDMA cnode7-013:736481:744977 [0] NCCL INFO Channel 06/0 : 0[0] -> 8[0] [receive] via NET/IB/0/GDRDMA cnode7-013:736481:744977 [0] NCCL INFO Channel 07/0 : 0[0] -> 8[0] [receive] via NET/IB/0/GDRDMA cnode7-013:736481:744977 [0] NCCL INFO Channel 10/0 : 0[0] -> 8[0] [receive] via NET/IB/0/GDRDMA cnode7-013:736481:744977 [0] NCCL INFO Channel 11/0 : 0[0] -> 8[0] [receive] via NET/IB/0/GDRDMA cnode7-013:736481:744977 [0] NCCL INFO Channel 12/0 : 0[0] -> 8[0] [receive] via NET/IB/0/GDRDMA cnode7-013:736481:744977 [0] NCCL INFO Channel 13/0 : 0[0] -> 8[0] [receive] via NET/IB/0/GDRDMA cnode7-013:736481:744977 [0] NCCL INFO Channel 14/0 : 0[0] -> 8[0] [receive] via NET/IB/0/GDRDMA cnode7-013:736481:744977 [0] NCCL INFO Channel 15/0 : 0[0] -> 8[0] [receive] via NET/IB/0/GDRDMA cnode7-013:736481:744977 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [send] via NET/IB/0/GDRDMA cnode7-013:736481:744977 [0] NCCL INFO Channel 02/0 : 8[0] -> 0[0] [send] via NET/IB/0/GDRDMA cnode7-013:736481:744977 [0] NCCL INFO Channel 03/0 : 8[0] -> 0[0] [send] via NET/IB/0/GDRDMA cnode7-013:736481:744977 [0] NCCL INFO Channel 04/0 : 8[0] -> 0[0] [send] via NET/IB/0/GDRDMA cnode7-013:736481:744977 [0] NCCL INFO Channel 05/0 : 8[0] -> 0[0] [send] via NET/IB/0/GDRDMA cnode7-013:736481:744977 [0] NCCL INFO Channel 06/0 : 8[0] -> 0[0] [send] via NET/IB/0/GDRDMA cnode7-013:736481:744977 [0] NCCL INFO Channel 07/0 : 8[0] -> 0[0] [send] via NET/IB/0/GDRDMA cnode7-013:736481:744977 [0] NCCL INFO Channel 09/0 : 8[0] -> 0[0] [send] via NET/IB/0/GDRDMA cnode7-013:736481:744977 [0] NCCL INFO Channel 10/0 : 8[0] -> 0[0] [send] via NET/IB/0/GDRDMA cnode7-013:736481:744977 [0] NCCL INFO Channel 11/0 : 8[0] -> 0[0] [send] via NET/IB/0/GDRDMA cnode7-013:736481:744977 [0] NCCL INFO Channel 12/0 : 8[0] -> 0[0] [send] via NET/IB/0/GDRDMA cnode7-013:736481:744977 [0] NCCL INFO Channel 13/0 : 8[0] -> 0[0] [send] via NET/IB/0/GDRDMA cnode7-013:736481:744977 [0] NCCL INFO Channel 14/0 : 8[0] -> 0[0] [send] via NET/IB/0/GDRDMA cnode7-013:736481:744977 [0] NCCL INFO Channel 15/0 : 8[0] -> 0[0] [send] via NET/IB/0/GDRDMA cnode7-013:736485:745370 [4] NCCL INFO Channel 00/0 : 4[4] -> 12[4] [receive] via NET/IB/4/GDRDMA cnode7-013:736485:745370 [4] NCCL INFO Channel 01/0 : 4[4] -> 12[4] [receive] via NET/IB/4/GDRDMA cnode7-013:736485:745370 [4] NCCL INFO Channel 02/0 : 4[4] -> 12[4] [receive] via NET/IB/4/GDRDMA cnode7-013:736485:745370 [4] NCCL INFO Channel 03/0 : 4[4] -> 12[4] [receive] via NET/IB/4/GDRDMA cnode7-013:736485:745370 [4] NCCL INFO Channel 06/0 : 4[4] -> 12[4] [receive] via NET/IB/4/GDRDMA cnode7-013:736485:745370 [4] NCCL INFO Channel 07/0 : 4[4] -> 12[4] [receive] via NET/IB/4/GDRDMA cnode7-013:736485:745370 [4] NCCL INFO Channel 08/0 : 4[4] -> 12[4] [receive] via NET/IB/4/GDRDMA cnode7-013:736485:745370 [4] NCCL INFO Channel 09/0 : 4[4] -> 12[4] [receive] via NET/IB/4/GDRDMA cnode7-013:736485:745370 [4] NCCL INFO Channel 10/0 : 4[4] -> 12[4] [receive] via NET/IB/4/GDRDMA cnode7-013:736485:745370 [4] NCCL INFO Channel 11/0 : 4[4] -> 12[4] [receive] via NET/IB/4/GDRDMA cnode7-013:736485:745370 [4] NCCL INFO Channel 14/0 : 4[4] -> 12[4] [receive] via NET/IB/4/GDRDMA cnode7-013:736485:745370 [4] NCCL INFO Channel 15/0 : 4[4] -> 12[4] [receive] via NET/IB/4/GDRDMA cnode7-013:736485:745370 [4] NCCL INFO Channel 00/0 : 12[4] -> 4[4] [send] via NET/IB/4/GDRDMA cnode7-013:736485:745370 [4] NCCL INFO Channel 01/0 : 12[4] -> 4[4] [send] via NET/IB/4/GDRDMA cnode7-013:736485:745370 [4] NCCL INFO Channel 02/0 : 12[4] -> 4[4] [send] via NET/IB/4/GDRDMA cnode7-013:736485:745370 [4] NCCL INFO Channel 03/0 : 12[4] -> 4[4] [send] via NET/IB/4/GDRDMA cnode7-013:736485:745370 [4] NCCL INFO Channel 05/0 : 12[4] -> 4[4] [send] via NET/IB/4/GDRDMA cnode7-013:736485:745370 [4] NCCL INFO Channel 06/0 : 12[4] -> 4[4] [send] via NET/IB/4/GDRDMA cnode7-013:736485:745370 [4] NCCL INFO Channel 07/0 : 12[4] -> 4[4] [send] via NET/IB/4/GDRDMA cnode7-013:736485:745370 [4] NCCL INFO Channel 08/0 : 12[4] -> 4[4] [send] via NET/IB/4/GDRDMA cnode7-013:736485:745370 [4] NCCL INFO Channel 09/0 : 12[4] -> 4[4] [send] via NET/IB/4/GDRDMA cnode7-013:736485:745370 [4] NCCL INFO Channel 10/0 : 12[4] -> 4[4] [send] via NET/IB/4/GDRDMA cnode7-013:736485:745370 [4] NCCL INFO Channel 11/0 : 12[4] -> 4[4] [send] via NET/IB/4/GDRDMA cnode7-013:736485:745370 [4] NCCL INFO Channel 13/0 : 12[4] -> 4[4] [send] via NET/IB/4/GDRDMA cnode7-013:736485:745370 [4] NCCL INFO Channel 14/0 : 12[4] -> 4[4] [send] via NET/IB/4/GDRDMA cnode7-013:736485:745370 [4] NCCL INFO Channel 15/0 : 12[4] -> 4[4] [send] via NET/IB/4/GDRDMA cnode7-013:736486:745644 [5] NCCL INFO Channel 00/0 : 5[5] -> 13[5] [receive] via NET/IB/6/GDRDMA cnode7-013:736486:745644 [5] NCCL INFO Channel 01/0 : 5[5] -> 13[5] [receive] via NET/IB/6/GDRDMA cnode7-013:736486:745644 [5] NCCL INFO Channel 02/0 : 5[5] -> 13[5] [receive] via NET/IB/6/GDRDMA cnode7-013:736486:745644 [5] NCCL INFO Channel 03/0 : 5[5] -> 13[5] [receive] via NET/IB/6/GDRDMA cnode7-013:736486:745644 [5] NCCL INFO Channel 06/0 : 5[5] -> 13[5] [receive] via NET/IB/6/GDRDMA cnode7-013:736486:745644 [5] NCCL INFO Channel 07/0 : 5[5] -> 13[5] [receive] via NET/IB/6/GDRDMA cnode7-013:736486:745644 [5] NCCL INFO Channel 08/0 : 5[5] -> 13[5] [receive] via NET/IB/6/GDRDMA cnode7-013:736486:745644 [5] NCCL INFO Channel 09/0 : 5[5] -> 13[5] [receive] via NET/IB/6/GDRDMA cnode7-013:736486:745644 [5] NCCL INFO Channel 10/0 : 5[5] -> 13[5] [receive] via NET/IB/6/GDRDMA cnode7-013:736486:745644 [5] NCCL INFO Channel 11/0 : 5[5] -> 13[5] [receive] via NET/IB/6/GDRDMA cnode7-013:736486:745644 [5] NCCL INFO Channel 14/0 : 5[5] -> 13[5] [receive] via NET/IB/6/GDRDMA cnode7-013:736486:745644 [5] NCCL INFO Channel 15/0 : 5[5] -> 13[5] [receive] via NET/IB/6/GDRDMA cnode7-013:736486:745644 [5] NCCL INFO Channel 00/0 : 13[5] -> 5[5] [send] via NET/IB/6/GDRDMA cnode7-013:736486:745644 [5] NCCL INFO Channel 01/0 : 13[5] -> 5[5] [send] via NET/IB/6/GDRDMA cnode7-013:736486:745644 [5] NCCL INFO Channel 02/0 : 13[5] -> 5[5] [send] via NET/IB/6/GDRDMA cnode7-013:736486:745644 [5] NCCL INFO Channel 03/0 : 13[5] -> 5[5] [send] via NET/IB/6/GDRDMA cnode7-013:736486:745644 [5] NCCL INFO Channel 04/0 : 13[5] -> 5[5] [send] via NET/IB/6/GDRDMA cnode7-013:736486:745644 [5] NCCL INFO Channel 06/0 : 13[5] -> 5[5] [send] via NET/IB/6/GDRDMA cnode7-013:736486:745644 [5] NCCL INFO Channel 07/0 : 13[5] -> 5[5] [send] via NET/IB/6/GDRDMA cnode7-013:736486:745644 [5] NCCL INFO Channel 08/0 : 13[5] -> 5[5] [send] via NET/IB/6/GDRDMA cnode7-013:736486:745644 [5] NCCL INFO Channel 09/0 : 13[5] -> 5[5] [send] via NET/IB/6/GDRDMA cnode7-013:736486:745644 [5] NCCL INFO Channel 10/0 : 13[5] -> 5[5] [send] via NET/IB/6/GDRDMA cnode7-013:736486:745644 [5] NCCL INFO Channel 11/0 : 13[5] -> 5[5] [send] via NET/IB/6/GDRDMA cnode7-013:736486:745644 [5] NCCL INFO Channel 12/0 : 13[5] -> 5[5] [send] via NET/IB/6/GDRDMA cnode7-013:736486:745644 [5] NCCL INFO Channel 14/0 : 13[5] -> 5[5] [send] via NET/IB/6/GDRDMA cnode7-013:736486:745644 [5] NCCL INFO Channel 15/0 : 13[5] -> 5[5] [send] via NET/IB/6/GDRDMA cnode7-012:1272012:1281554 [7] NCCL INFO Channel 00/0 : 15[7] -> 7[7] [receive] via NET/IB/8/GDRDMA cnode7-012:1272012:1281554 [7] NCCL INFO Channel 01/0 : 15[7] -> 7[7] [receive] via NET/IB/8/GDRDMA cnode7-012:1272012:1281554 [7] NCCL INFO Channel 02/0 : 15[7] -> 7[7] [receive] via NET/IB/8/GDRDMA cnode7-012:1272012:1281554 [7] NCCL INFO Channel 03/0 : 15[7] -> 7[7] [receive] via NET/IB/8/GDRDMA cnode7-012:1272012:1281554 [7] NCCL INFO Channel 04/0 : 15[7] -> 7[7] [receive] via NET/IB/8/GDRDMA cnode7-012:1272012:1281554 [7] NCCL INFO Channel 05/0 : 15[7] -> 7[7] [receive] via NET/IB/8/GDRDMA cnode7-012:1272012:1281554 [7] NCCL INFO Channel 06/0 : 15[7] -> 7[7] [receive] via NET/IB/8/GDRDMA cnode7-012:1272012:1281554 [7] NCCL INFO Channel 08/0 : 15[7] -> 7[7] [receive] via NET/IB/8/GDRDMA cnode7-012:1272012:1281554 [7] NCCL INFO Channel 09/0 : 15[7] -> 7[7] [receive] via NET/IB/8/GDRDMA cnode7-012:1272012:1281554 [7] NCCL INFO Channel 10/0 : 15[7] -> 7[7] [receive] via NET/IB/8/GDRDMA cnode7-012:1272012:1281554 [7] NCCL INFO Channel 11/0 : 15[7] -> 7[7] [receive] via NET/IB/8/GDRDMA cnode7-012:1272012:1281554 [7] NCCL INFO Channel 12/0 : 15[7] -> 7[7] [receive] via NET/IB/8/GDRDMA cnode7-012:1272012:1281554 [7] NCCL INFO Channel 13/0 : 15[7] -> 7[7] [receive] via NET/IB/8/GDRDMA cnode7-012:1272012:1281554 [7] NCCL INFO Channel 14/0 : 15[7] -> 7[7] [receive] via NET/IB/8/GDRDMA cnode7-012:1272012:1281554 [7] NCCL INFO Channel 00/0 : 7[7] -> 15[7] [send] via NET/IB/8/GDRDMA cnode7-012:1272012:1281554 [7] NCCL INFO Channel 01/0 : 7[7] -> 15[7] [send] via NET/IB/8/GDRDMA cnode7-012:1272012:1281554 [7] NCCL INFO Channel 02/0 : 7[7] -> 15[7] [send] via NET/IB/8/GDRDMA cnode7-012:1272012:1281554 [7] NCCL INFO Channel 03/0 : 7[7] -> 15[7] [send] via NET/IB/8/GDRDMA cnode7-012:1272012:1281554 [7] NCCL INFO Channel 04/0 : 7[7] -> 15[7] [send] via NET/IB/8/GDRDMA cnode7-012:1272012:1281554 [7] NCCL INFO Channel 05/0 : 7[7] -> 15[7] [send] via NET/IB/8/GDRDMA cnode7-012:1272012:1281554 [7] NCCL INFO Channel 08/0 : 7[7] -> 15[7] [send] via NET/IB/8/GDRDMA cnode7-012:1272012:1281554 [7] NCCL INFO Channel 09/0 : 7[7] -> 15[7] [send] via NET/IB/8/GDRDMA cnode7-012:1272012:1281554 [7] NCCL INFO Channel 10/0 : 7[7] -> 15[7] [send] via NET/IB/8/GDRDMA cnode7-012:1272012:1281554 [7] NCCL INFO Channel 11/0 : 7[7] -> 15[7] [send] via NET/IB/8/GDRDMA cnode7-012:1272012:1281554 [7] NCCL INFO Channel 12/0 : 7[7] -> 15[7] [send] via NET/IB/8/GDRDMA cnode7-012:1272012:1281554 [7] NCCL INFO Channel 13/0 : 7[7] -> 15[7] [send] via NET/IB/8/GDRDMA cnode7-012:1272009:1281281 [4] NCCL INFO Channel 00/0 : 12[4] -> 4[4] [receive] via NET/IB/4/GDRDMA cnode7-012:1272009:1281281 [4] NCCL INFO Channel 01/0 : 12[4] -> 4[4] [receive] via NET/IB/4/GDRDMA cnode7-012:1272009:1281281 [4] NCCL INFO Channel 02/0 : 12[4] -> 4[4] [receive] via NET/IB/4/GDRDMA cnode7-012:1272009:1281281 [4] NCCL INFO Channel 03/0 : 12[4] -> 4[4] [receive] via NET/IB/4/GDRDMA cnode7-012:1272009:1281281 [4] NCCL INFO Channel 05/0 : 12[4] -> 4[4] [receive] via NET/IB/4/GDRDMA cnode7-012:1272009:1281281 [4] NCCL INFO Channel 06/0 : 12[4] -> 4[4] [receive] via NET/IB/4/GDRDMA cnode7-012:1272009:1281281 [4] NCCL INFO Channel 07/0 : 12[4] -> 4[4] [receive] via NET/IB/4/GDRDMA cnode7-012:1272009:1281281 [4] NCCL INFO Channel 08/0 : 12[4] -> 4[4] [receive] via NET/IB/4/GDRDMA cnode7-012:1272009:1281281 [4] NCCL INFO Channel 09/0 : 12[4] -> 4[4] [receive] via NET/IB/4/GDRDMA cnode7-012:1272009:1281281 [4] NCCL INFO Channel 10/0 : 12[4] -> 4[4] [receive] via NET/IB/4/GDRDMA cnode7-012:1272009:1281281 [4] NCCL INFO Channel 11/0 : 12[4] -> 4[4] [receive] via NET/IB/4/GDRDMA cnode7-012:1272009:1281281 [4] NCCL INFO Channel 13/0 : 12[4] -> 4[4] [receive] via NET/IB/4/GDRDMA cnode7-012:1272009:1281281 [4] NCCL INFO Channel 14/0 : 12[4] -> 4[4] [receive] via NET/IB/4/GDRDMA cnode7-012:1272009:1281281 [4] NCCL INFO Channel 15/0 : 12[4] -> 4[4] [receive] via NET/IB/4/GDRDMA cnode7-012:1272009:1281281 [4] NCCL INFO Channel 00/0 : 4[4] -> 12[4] [send] via NET/IB/4/GDRDMA cnode7-012:1272009:1281281 [4] NCCL INFO Channel 01/0 : 4[4] -> 12[4] [send] via NET/IB/4/GDRDMA cnode7-012:1272009:1281281 [4] NCCL INFO Channel 02/0 : 4[4] -> 12[4] [send] via NET/IB/4/GDRDMA cnode7-012:1272009:1281281 [4] NCCL INFO Channel 03/0 : 4[4] -> 12[4] [send] via NET/IB/4/GDRDMA cnode7-012:1272009:1281281 [4] NCCL INFO Channel 06/0 : 4[4] -> 12[4] [send] via NET/IB/4/GDRDMA cnode7-012:1272009:1281281 [4] NCCL INFO Channel 07/0 : 4[4] -> 12[4] [send] via NET/IB/4/GDRDMA cnode7-012:1272009:1281281 [4] NCCL INFO Channel 08/0 : 4[4] -> 12[4] [send] via NET/IB/4/GDRDMA cnode7-012:1272009:1281281 [4] NCCL INFO Channel 09/0 : 4[4] -> 12[4] [send] via NET/IB/4/GDRDMA cnode7-012:1272009:1281281 [4] NCCL INFO Channel 10/0 : 4[4] -> 12[4] [send] via NET/IB/4/GDRDMA cnode7-012:1272009:1281281 [4] NCCL INFO Channel 11/0 : 4[4] -> 12[4] [send] via NET/IB/4/GDRDMA cnode7-012:1272009:1281281 [4] NCCL INFO Channel 14/0 : 4[4] -> 12[4] [send] via NET/IB/4/GDRDMA cnode7-012:1272009:1281281 [4] NCCL INFO Channel 15/0 : 4[4] -> 12[4] [send] via NET/IB/4/GDRDMA cnode7-012:1272006:1281620 [1] NCCL INFO Channel 00/0 : 9[1] -> 1[1] [receive] via NET/IB/1/GDRDMA cnode7-012:1272006:1281620 [1] NCCL INFO Channel 02/0 : 9[1] -> 1[1] [receive] via NET/IB/1/GDRDMA cnode7-012:1272006:1281620 [1] NCCL INFO Channel 03/0 : 9[1] -> 1[1] [receive] via NET/IB/1/GDRDMA cnode7-012:1272006:1281620 [1] NCCL INFO Channel 04/0 : 9[1] -> 1[1] [receive] via NET/IB/1/GDRDMA cnode7-012:1272006:1281620 [1] NCCL INFO Channel 05/0 : 9[1] -> 1[1] [receive] via NET/IB/1/GDRDMA cnode7-012:1272006:1281620 [1] NCCL INFO Channel 06/0 : 9[1] -> 1[1] [receive] via NET/IB/1/GDRDMA cnode7-012:1272006:1281620 [1] NCCL INFO Channel 07/0 : 9[1] -> 1[1] [receive] via NET/IB/1/GDRDMA cnode7-012:1272006:1281620 [1] NCCL INFO Channel 08/0 : 9[1] -> 1[1] [receive] via NET/IB/1/GDRDMA cnode7-012:1272006:1281620 [1] NCCL INFO Channel 10/0 : 9[1] -> 1[1] [receive] via NET/IB/1/GDRDMA cnode7-012:1272006:1281620 [1] NCCL INFO Channel 11/0 : 9[1] -> 1[1] [receive] via NET/IB/1/GDRDMA cnode7-012:1272006:1281620 [1] NCCL INFO Channel 12/0 : 9[1] -> 1[1] [receive] via NET/IB/1/GDRDMA cnode7-012:1272006:1281620 [1] NCCL INFO Channel 13/0 : 9[1] -> 1[1] [receive] via NET/IB/1/GDRDMA cnode7-012:1272006:1281620 [1] NCCL INFO Channel 14/0 : 9[1] -> 1[1] [receive] via NET/IB/1/GDRDMA cnode7-012:1272006:1281620 [1] NCCL INFO Channel 15/0 : 9[1] -> 1[1] [receive] via NET/IB/1/GDRDMA cnode7-012:1272006:1281620 [1] NCCL INFO Channel 02/0 : 1[1] -> 9[1] [send] via NET/IB/1/GDRDMA cnode7-012:1272006:1281620 [1] NCCL INFO Channel 03/0 : 1[1] -> 9[1] [send] via NET/IB/1/GDRDMA cnode7-012:1272006:1281620 [1] NCCL INFO Channel 04/0 : 1[1] -> 9[1] [send] via NET/IB/1/GDRDMA cnode7-012:1272006:1281620 [1] NCCL INFO Channel 05/0 : 1[1] -> 9[1] [send] via NET/IB/1/GDRDMA cnode7-012:1272006:1281620 [1] NCCL INFO Channel 06/0 : 1[1] -> 9[1] [send] via NET/IB/1/GDRDMA cnode7-012:1272006:1281620 [1] NCCL INFO Channel 07/0 : 1[1] -> 9[1] [send] via NET/IB/1/GDRDMA cnode7-012:1272006:1281620 [1] NCCL INFO Channel 10/0 : 1[1] -> 9[1] [send] via NET/IB/1/GDRDMA cnode7-012:1272006:1281620 [1] NCCL INFO Channel 11/0 : 1[1] -> 9[1] [send] via NET/IB/1/GDRDMA cnode7-012:1272006:1281620 [1] NCCL INFO Channel 12/0 : 1[1] -> 9[1] [send] via NET/IB/1/GDRDMA cnode7-012:1272006:1281620 [1] NCCL INFO Channel 13/0 : 1[1] -> 9[1] [send] via NET/IB/1/GDRDMA cnode7-012:1272006:1281620 [1] NCCL INFO Channel 14/0 : 1[1] -> 9[1] [send] via NET/IB/1/GDRDMA cnode7-012:1272006:1281620 [1] NCCL INFO Channel 15/0 : 1[1] -> 9[1] [send] via NET/IB/1/GDRDMA cnode7-012:1272011:1281645 [6] NCCL INFO Channel 00/0 : 14[6] -> 6[6] [receive] via NET/IB/7/GDRDMA cnode7-012:1272011:1281645 [6] NCCL INFO Channel 01/0 : 14[6] -> 6[6] [receive] via NET/IB/7/GDRDMA cnode7-012:1272011:1281645 [6] NCCL INFO Channel 02/0 : 14[6] -> 6[6] [receive] via NET/IB/7/GDRDMA cnode7-012:1272011:1281645 [6] NCCL INFO Channel 03/0 : 14[6] -> 6[6] [receive] via NET/IB/7/GDRDMA cnode7-012:1272011:1281645 [6] NCCL INFO Channel 04/0 : 14[6] -> 6[6] [receive] via NET/IB/7/GDRDMA cnode7-012:1272011:1281645 [6] NCCL INFO Channel 05/0 : 14[6] -> 6[6] [receive] via NET/IB/7/GDRDMA cnode7-012:1272011:1281645 [6] NCCL INFO Channel 07/0 : 14[6] -> 6[6] [receive] via NET/IB/7/GDRDMA cnode7-012:1272011:1281645 [6] NCCL INFO Channel 08/0 : 14[6] -> 6[6] [receive] via NET/IB/7/GDRDMA cnode7-012:1272011:1281645 [6] NCCL INFO Channel 09/0 : 14[6] -> 6[6] [receive] via NET/IB/7/GDRDMA cnode7-012:1272011:1281645 [6] NCCL INFO Channel 10/0 : 14[6] -> 6[6] [receive] via NET/IB/7/GDRDMA cnode7-012:1272011:1281645 [6] NCCL INFO Channel 11/0 : 14[6] -> 6[6] [receive] via NET/IB/7/GDRDMA cnode7-012:1272011:1281645 [6] NCCL INFO Channel 12/0 : 14[6] -> 6[6] [receive] via NET/IB/7/GDRDMA cnode7-012:1272011:1281645 [6] NCCL INFO Channel 13/0 : 14[6] -> 6[6] [receive] via NET/IB/7/GDRDMA cnode7-012:1272011:1281645 [6] NCCL INFO Channel 15/0 : 14[6] -> 6[6] [receive] via NET/IB/7/GDRDMA cnode7-012:1272011:1281645 [6] NCCL INFO Channel 00/0 : 6[6] -> 14[6] [send] via NET/IB/7/GDRDMA cnode7-012:1272011:1281645 [6] NCCL INFO Channel 01/0 : 6[6] -> 14[6] [send] via NET/IB/7/GDRDMA cnode7-012:1272011:1281645 [6] NCCL INFO Channel 02/0 : 6[6] -> 14[6] [send] via NET/IB/7/GDRDMA cnode7-012:1272011:1281645 [6] NCCL INFO Channel 03/0 : 6[6] -> 14[6] [send] via NET/IB/7/GDRDMA cnode7-012:1272011:1281645 [6] NCCL INFO Channel 04/0 : 6[6] -> 14[6] [send] via NET/IB/7/GDRDMA cnode7-012:1272011:1281645 [6] NCCL INFO Channel 05/0 : 6[6] -> 14[6] [send] via NET/IB/7/GDRDMA cnode7-012:1272011:1281645 [6] NCCL INFO Channel 08/0 : 6[6] -> 14[6] [send] via NET/IB/7/GDRDMA cnode7-012:1272011:1281645 [6] NCCL INFO Channel 09/0 : 6[6] -> 14[6] [send] via NET/IB/7/GDRDMA cnode7-012:1272011:1281645 [6] NCCL INFO Channel 10/0 : 6[6] -> 14[6] [send] via NET/IB/7/GDRDMA cnode7-012:1272011:1281645 [6] NCCL INFO Channel 11/0 : 6[6] -> 14[6] [send] via NET/IB/7/GDRDMA cnode7-012:1272011:1281645 [6] NCCL INFO Channel 12/0 : 6[6] -> 14[6] [send] via NET/IB/7/GDRDMA cnode7-012:1272011:1281645 [6] NCCL INFO Channel 13/0 : 6[6] -> 14[6] [send] via NET/IB/7/GDRDMA cnode7-013:736483:745475 [2] NCCL INFO Channel 03/0 : 10[2] -> 2[2] [send] via NET/IB/2/GDRDMA cnode7-013:736483:745475 [2] NCCL INFO Channel 04/0 : 10[2] -> 2[2] [send] via NET/IB/2/GDRDMA cnode7-013:736483:745475 [2] NCCL INFO Channel 05/0 : 10[2] -> 2[2] [send] via NET/IB/2/GDRDMA cnode7-013:736483:745475 [2] NCCL INFO Channel 06/0 : 10[2] -> 2[2] [send] via NET/IB/2/GDRDMA cnode7-013:736483:745475 [2] NCCL INFO Channel 07/0 : 10[2] -> 2[2] [send] via NET/IB/2/GDRDMA cnode7-013:736483:745475 [2] NCCL INFO Channel 08/0 : 10[2] -> 2[2] [send] via NET/IB/2/GDRDMA cnode7-013:736483:745475 [2] NCCL INFO Channel 09/0 : 10[2] -> 2[2] [send] via NET/IB/2/GDRDMA cnode7-013:736483:745475 [2] NCCL INFO Channel 11/0 : 10[2] -> 2[2] [send] via NET/IB/2/GDRDMA cnode7-013:736483:745475 [2] NCCL INFO Channel 12/0 : 10[2] -> 2[2] [send] via NET/IB/2/GDRDMA cnode7-013:736483:745475 [2] NCCL INFO Channel 13/0 : 10[2] -> 2[2] [send] via NET/IB/2/GDRDMA cnode7-013:736483:745475 [2] NCCL INFO Channel 14/0 : 10[2] -> 2[2] [send] via NET/IB/2/GDRDMA cnode7-013:736483:745475 [2] NCCL INFO Channel 15/0 : 10[2] -> 2[2] [send] via NET/IB/2/GDRDMA cnode7-012:1272010:1281060 [5] NCCL INFO Channel 15/0 : 5[5] -> 13[5] [send] via NET/IB/6/GDRDMA cnode7-013:736483:745475 [2] NCCL INFO Connected NVLS tree cnode7-013:736483:745475 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512 cnode7-013:736483:745475 [2] NCCL INFO 16 coll channels, 0 collnet channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer cnode7-013:736482:746397 [1] NCCL INFO Connected NVLS tree cnode7-013:736482:746397 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512 cnode7-013:736482:746397 [1] NCCL INFO 16 coll channels, 0 collnet channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer cnode7-012:1272007:1281061 [2] NCCL INFO Connected NVLS tree cnode7-012:1272007:1281061 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512 cnode7-012:1272007:1281061 [2] NCCL INFO 16 coll channels, 0 collnet channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer cnode7-013:736481:744977 [0] NCCL INFO Connected NVLS tree cnode7-013:736481:744977 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512 cnode7-013:736481:744977 [0] NCCL INFO 16 coll channels, 0 collnet channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer cnode7-013:736486:745644 [5] NCCL INFO Connected NVLS tree cnode7-013:736486:745644 [5] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512 cnode7-013:736486:745644 [5] NCCL INFO 16 coll channels, 0 collnet channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer cnode7-012:1272012:1281554 [7] NCCL INFO Connected NVLS tree cnode7-012:1272012:1281554 [7] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512 cnode7-012:1272012:1281554 [7] NCCL INFO 16 coll channels, 0 collnet channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer cnode7-013:736488:746041 [7] NCCL INFO Connected NVLS tree cnode7-013:736488:746041 [7] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512 cnode7-013:736488:746041 [7] NCCL INFO 16 coll channels, 0 collnet channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer cnode7-013:736487:746330 [6] NCCL INFO Connected NVLS tree cnode7-013:736487:746330 [6] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512 cnode7-013:736487:746330 [6] NCCL INFO 16 coll channels, 0 collnet channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer cnode7-013:736484:745381 [3] NCCL INFO Connected NVLS tree cnode7-013:736484:745381 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512 cnode7-013:736484:745381 [3] NCCL INFO 16 coll channels, 0 collnet channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer cnode7-012:1272006:1281620 [1] NCCL INFO Connected NVLS tree cnode7-012:1272006:1281620 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512 cnode7-012:1272006:1281620 [1] NCCL INFO 16 coll channels, 0 collnet channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer cnode7-012:1272011:1281645 [6] NCCL INFO Connected NVLS tree cnode7-012:1272011:1281645 [6] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512 cnode7-012:1272011:1281645 [6] NCCL INFO 16 coll channels, 0 collnet channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer cnode7-012:1272005:1281059 [0] NCCL INFO Connected NVLS tree cnode7-012:1272005:1281059 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512 cnode7-012:1272005:1281059 [0] NCCL INFO 16 coll channels, 0 collnet channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer cnode7-013:736485:745370 [4] NCCL INFO Connected NVLS tree cnode7-013:736485:745370 [4] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512 cnode7-013:736485:745370 [4] NCCL INFO 16 coll channels, 0 collnet channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer cnode7-013:736483:745475 [2] NCCL INFO comm 0x5555cf9ca470 rank 10 nranks 16 cudaDev 2 nvmlDev 2 busId 52000 commId 0x17e756722888e895 - Init COMPLETE cnode7-012:1272010:1281060 [5] NCCL INFO Connected NVLS tree cnode7-012:1272010:1281060 [5] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512 cnode7-012:1272010:1281060 [5] NCCL INFO 16 coll channels, 0 collnet channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer cnode7-012:1272008:1281624 [3] NCCL INFO Connected NVLS tree cnode7-012:1272008:1281624 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512 cnode7-012:1272008:1281624 [3] NCCL INFO 16 coll channels, 0 collnet channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer cnode7-012:1272009:1281281 [4] NCCL INFO Connected NVLS tree cnode7-012:1272009:1281281 [4] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512 cnode7-012:1272009:1281281 [4] NCCL INFO 16 coll channels, 0 collnet channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer cnode7-013:736488:746041 [7] NCCL INFO comm 0x5555cf9aad40 rank 15 nranks 16 cudaDev 7 nvmlDev 7 busId df000 commId 0x17e756722888e895 - Init COMPLETE cnode7-012:1272008:1281624 [3] NCCL INFO comm 0x5555e39b2180 rank 3 nranks 16 cudaDev 3 nvmlDev 3 busId 61000 commId 0x17e756722888e895 - Init COMPLETE cnode7-013:736482:746397 [1] NCCL INFO comm 0x5555cf9ca2c0 rank 9 nranks 16 cudaDev 1 nvmlDev 1 busId 43000 commId 0x17e756722888e895 - Init COMPLETE cnode7-013:736487:746330 [6] NCCL INFO comm 0x5555cf9ae150 rank 14 nranks 16 cudaDev 6 nvmlDev 6 busId d1000 commId 0x17e756722888e895 - Init COMPLETE cnode7-013:736486:745644 [5] NCCL INFO comm 0x5555cd22bfa0 rank 13 nranks 16 cudaDev 5 nvmlDev 5 busId c3000 commId 0x17e756722888e895 - Init COMPLETE cnode7-013:736485:745370 [4] NCCL INFO comm 0x5555cf9ca4a0 rank 12 nranks 16 cudaDev 4 nvmlDev 4 busId 9d000 commId 0x17e756722888e895 - Init COMPLETE cnode7-013:736481:744977 [0] NCCL INFO comm 0x5555c59f0f70 rank 8 nranks 16 cudaDev 0 nvmlDev 0 busId 1b000 commId 0x17e756722888e895 - Init COMPLETE cnode7-013:736484:745381 [3] NCCL INFO comm 0x5555cd20c9e0 rank 11 nranks 16 cudaDev 3 nvmlDev 3 busId 61000 commId 0x17e756722888e895 - Init COMPLETE cnode7-012:1272012:1281554 [7] NCCL INFO comm 0x5555cf991e20 rank 7 nranks 16 cudaDev 7 nvmlDev 7 busId df000 commId 0x17e756722888e895 - Init COMPLETE cnode7-012:1272006:1281620 [1] NCCL INFO comm 0x5555cf9b5e30 rank 1 nranks 16 cudaDev 1 nvmlDev 1 busId 43000 commId 0x17e756722888e895 - Init COMPLETE cnode7-012:1272007:1281061 [2] NCCL INFO comm 0x5555e39b9f60 rank 2 nranks 16 cudaDev 2 nvmlDev 2 busId 52000 commId 0x17e756722888e895 - Init COMPLETE cnode7-012:1272011:1281645 [6] NCCL INFO comm 0x5555cf994130 rank 6 nranks 16 cudaDev 6 nvmlDev 6 busId d1000 commId 0x17e756722888e895 - Init COMPLETE cnode7-012:1272009:1281281 [4] NCCL INFO comm 0x5555cd223a50 rank 4 nranks 16 cudaDev 4 nvmlDev 4 busId 9d000 commId 0x17e756722888e895 - Init COMPLETE cnode7-012:1272010:1281060 [5] NCCL INFO comm 0x5555e39cd100 rank 5 nranks 16 cudaDev 5 nvmlDev 5 busId c3000 commId 0x17e756722888e895 - Init COMPLETE cnode7-012:1272005:1281059 [0] NCCL INFO comm 0x5555c2f5fad0 rank 0 nranks 16 cudaDev 0 nvmlDev 0 busId 1b000 commId 0x17e756722888e895 - Init COMPLETE [cnode7-012:0/16] 2024-12-05 16:39:13,921 (deepspeed_trainer:228) INFO: 13epoch:train:1-100batch: iter_time=2.374, loss_ctc=84.201, loss_att=70.775, acc=0.670, loss=74.778, grad_norm=4.009, loss_scale=1.000, learning_rate=1.155e-04, step_time=0.601 [cnode7-012:0/16] 2024-12-05 16:39:49,932 (deepspeed_trainer:228) INFO: 13epoch:train:101-200batch: iter_time=1.295e-04, loss_ctc=80.346, loss_att=60.906, acc=0.654, loss=66.732, grad_norm=4.175, loss_scale=1.000, learning_rate=1.154e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 16:40:26,025 (deepspeed_trainer:228) INFO: 13epoch:train:201-300batch: iter_time=1.311e-04, loss_ctc=94.762, loss_att=79.626, acc=0.641, loss=84.164, grad_norm=4.986, loss_scale=1.000, learning_rate=1.154e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 16:41:01,770 (deepspeed_trainer:228) INFO: 13epoch:train:301-400batch: iter_time=1.435e-04, loss_ctc=81.458, loss_att=61.775, acc=0.654, loss=67.670, grad_norm=4.054, loss_scale=1.000, learning_rate=1.154e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-05 16:41:37,597 (deepspeed_trainer:228) INFO: 13epoch:train:401-500batch: iter_time=1.425e-04, loss_ctc=78.076, loss_att=59.976, acc=0.669, loss=65.397, grad_norm=3.998, loss_scale=1.000, learning_rate=1.153e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 16:42:13,511 (deepspeed_trainer:228) INFO: 13epoch:train:501-600batch: iter_time=1.376e-04, loss_ctc=86.227, loss_att=70.167, acc=0.658, loss=74.973, grad_norm=4.140, loss_scale=1.000, learning_rate=1.153e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 16:42:49,546 (deepspeed_trainer:228) INFO: 13epoch:train:601-700batch: iter_time=1.330e-04, loss_ctc=90.834, loss_att=64.583, acc=0.660, loss=72.489, grad_norm=4.372, loss_scale=1.000, learning_rate=1.153e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 16:43:25,688 (deepspeed_trainer:228) INFO: 13epoch:train:701-800batch: iter_time=1.310e-04, loss_ctc=89.021, loss_att=74.506, acc=0.658, loss=78.841, grad_norm=4.324, loss_scale=1.000, learning_rate=1.152e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 16:44:01,826 (deepspeed_trainer:228) INFO: 13epoch:train:801-900batch: iter_time=1.278e-04, loss_ctc=99.424, loss_att=76.528, acc=0.658, loss=83.392, grad_norm=4.528, loss_scale=1.000, learning_rate=1.152e-04, step_time=0.361 [2024-12-05 16:44:37,942] [INFO] [logging.py:129:log_dist] [Rank 0] step=181000, skipped=0, lr=[np.float64(0.00011515031589559348)], mom=[[0.9, 0.98]] [2024-12-05 16:44:37,943] [INFO] [timer.py:264:stop] epoch=0/micro_step=1000/global_step=1000, RunningAvgSamplesPerSec=45.517989763860406, CurrSamplesPerSec=45.0012340218621, MemAllocated=1.77GB, MaxMemAllocated=16.04GB [cnode7-012:0/16] 2024-12-05 16:44:37,945 (deepspeed_trainer:228) INFO: 13epoch:train:901-1000batch: iter_time=1.281e-04, loss_ctc=81.803, loss_att=66.924, acc=0.660, loss=71.421, grad_norm=3.881, loss_scale=1.000, learning_rate=1.152e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 16:45:13,811 (deepspeed_trainer:228) INFO: 13epoch:train:1001-1100batch: iter_time=1.269e-04, loss_ctc=88.401, loss_att=64.548, acc=0.680, loss=71.724, grad_norm=3.964, loss_scale=1.000, learning_rate=1.151e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 16:45:49,592 (deepspeed_trainer:228) INFO: 13epoch:train:1101-1200batch: iter_time=1.223e-04, loss_ctc=77.579, loss_att=62.101, acc=0.665, loss=66.758, grad_norm=4.272, loss_scale=1.000, learning_rate=1.151e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-05 16:46:25,621 (deepspeed_trainer:228) INFO: 13epoch:train:1201-1300batch: iter_time=1.234e-04, loss_ctc=103.846, loss_att=79.198, acc=0.651, loss=86.588, grad_norm=4.918, loss_scale=1.000, learning_rate=1.151e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 16:47:01,831 (deepspeed_trainer:228) INFO: 13epoch:train:1301-1400batch: iter_time=1.202e-04, loss_ctc=90.265, loss_att=71.367, acc=0.644, loss=77.031, grad_norm=4.832, loss_scale=1.000, learning_rate=1.150e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 16:47:38,055 (deepspeed_trainer:228) INFO: 13epoch:train:1401-1500batch: iter_time=1.193e-04, loss_ctc=76.668, loss_att=60.353, acc=0.660, loss=65.256, grad_norm=3.765, loss_scale=1.000, learning_rate=1.150e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 16:48:14,037 (deepspeed_trainer:228) INFO: 13epoch:train:1501-1600batch: iter_time=1.253e-04, loss_ctc=85.241, loss_att=69.798, acc=0.676, loss=74.455, grad_norm=3.878, loss_scale=1.000, learning_rate=1.150e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 16:48:50,312 (deepspeed_trainer:228) INFO: 13epoch:train:1601-1700batch: iter_time=1.288e-04, loss_ctc=93.094, loss_att=72.414, acc=0.663, loss=78.633, grad_norm=4.152, loss_scale=1.000, learning_rate=1.149e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 16:49:26,546 (deepspeed_trainer:228) INFO: 13epoch:train:1701-1800batch: iter_time=1.232e-04, loss_ctc=88.065, loss_att=66.881, acc=0.667, loss=73.214, grad_norm=4.088, loss_scale=1.000, learning_rate=1.149e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 16:49:58,083 (multiple_iter_factory:32) INFO: Building 1th iter-factory... [cnode7-012:0/16] 2024-12-05 16:50:24,312 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-05 16:50:40,035 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.7", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.7", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.7", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.7", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-05 16:50:40,036 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.7, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-05 16:50:40,038 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-05 16:51:03,098] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:51:03,598] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:51:05,050] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:51:06,147] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:51:07,361] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:51:07,411] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:51:07,453] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:51:07,778] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:51:07,970] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:51:08,140] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:51:08,166] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:51:08,317] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:51:08,268] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:51:08,414] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:51:08,813] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:51:13,135] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:51:50,922] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:51:51,035] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:51:52,989] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:51:54,136] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:51:55,134] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:51:55,486] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:51:55,940] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:51:56,052] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:51:56,412] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:51:56,601] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:51:56,870] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:51:57,412] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:51:57,442] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:51:57,739] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:51:59,410] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:52:00,194] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:52:37,112] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:52:37,378] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:52:40,373] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:52:42,008] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:52:42,224] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:52:42,998] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:52:42,971] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:52:43,810] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:52:44,759] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:52:44,797] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:52:45,326] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:52:45,235] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:52:45,714] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:52:47,100] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:52:48,319] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:52:49,830] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:53:22,884] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:53:23,856] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:53:29,708] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:53:30,162] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:53:30,957] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:53:31,253] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:53:31,528] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:53:31,795] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:53:32,580] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:53:32,633] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:53:32,926] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:53:33,356] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:53:33,586] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:53:34,137] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:53:38,153] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 16:53:39,204] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-05 16:54:13,346 (deepspeed_trainer:228) INFO: 13epoch:train:1801-1900batch: iter_time=2.458, loss_ctc=73.752, loss_att=54.757, acc=0.666, loss=60.443, grad_norm=4.186, loss_scale=1.000, learning_rate=1.149e-04, step_time=0.409 [2024-12-05 16:54:49,569] [INFO] [logging.py:129:log_dist] [Rank 0] step=182000, skipped=0, lr=[np.float64(0.0001148335348763321)], mom=[[0.9, 0.98]] [2024-12-05 16:54:49,569] [INFO] [timer.py:264:stop] epoch=0/micro_step=2000/global_step=2000, RunningAvgSamplesPerSec=45.19380159131919, CurrSamplesPerSec=46.83945532336351, MemAllocated=1.77GB, MaxMemAllocated=16.04GB [cnode7-012:0/16] 2024-12-05 16:54:49,571 (deepspeed_trainer:228) INFO: 13epoch:train:1901-2000batch: iter_time=1.036e-04, loss_ctc=82.724, loss_att=65.328, acc=0.678, loss=70.529, grad_norm=3.867, loss_scale=1.000, learning_rate=1.148e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 16:55:25,429 (deepspeed_trainer:228) INFO: 13epoch:train:2001-2100batch: iter_time=1.050e-04, loss_ctc=83.005, loss_att=64.891, acc=0.656, loss=70.330, grad_norm=4.008, loss_scale=1.000, learning_rate=1.148e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 16:56:00,847 (deepspeed_trainer:228) INFO: 13epoch:train:2101-2200batch: iter_time=1.117e-04, loss_ctc=88.073, loss_att=68.400, acc=0.640, loss=74.275, grad_norm=4.849, loss_scale=1.000, learning_rate=1.148e-04, step_time=0.354 [cnode7-012:0/16] 2024-12-05 16:56:36,104 (deepspeed_trainer:228) INFO: 13epoch:train:2201-2300batch: iter_time=1.112e-04, loss_ctc=76.494, loss_att=56.901, acc=0.660, loss=62.791, grad_norm=3.817, loss_scale=1.000, learning_rate=1.148e-04, step_time=0.352 [cnode7-012:0/16] 2024-12-05 16:57:11,673 (deepspeed_trainer:228) INFO: 13epoch:train:2301-2400batch: iter_time=1.089e-04, loss_ctc=77.179, loss_att=59.414, acc=0.662, loss=64.767, grad_norm=3.960, loss_scale=1.000, learning_rate=1.147e-04, step_time=0.356 [cnode7-012:0/16] 2024-12-05 16:57:47,306 (deepspeed_trainer:228) INFO: 13epoch:train:2401-2500batch: iter_time=1.088e-04, loss_ctc=87.730, loss_att=69.022, acc=0.658, loss=74.620, grad_norm=4.181, loss_scale=1.000, learning_rate=1.147e-04, step_time=0.356 [cnode7-012:0/16] 2024-12-05 16:58:22,957 (deepspeed_trainer:228) INFO: 13epoch:train:2501-2600batch: iter_time=1.089e-04, loss_ctc=90.468, loss_att=63.701, acc=0.664, loss=71.764, grad_norm=4.334, loss_scale=1.000, learning_rate=1.147e-04, step_time=0.356 [cnode7-012:0/16] 2024-12-05 16:58:58,523 (deepspeed_trainer:228) INFO: 13epoch:train:2601-2700batch: iter_time=1.105e-04, loss_ctc=81.581, loss_att=69.514, acc=0.654, loss=73.196, grad_norm=4.237, loss_scale=1.000, learning_rate=1.146e-04, step_time=0.355 [cnode7-012:0/16] 2024-12-05 16:59:34,408 (deepspeed_trainer:228) INFO: 13epoch:train:2701-2800batch: iter_time=1.114e-04, loss_ctc=102.022, loss_att=80.186, acc=0.651, loss=86.752, grad_norm=4.936, loss_scale=1.000, learning_rate=1.146e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 17:00:09,840 (deepspeed_trainer:228) INFO: 13epoch:train:2801-2900batch: iter_time=1.121e-04, loss_ctc=80.622, loss_att=62.773, acc=0.669, loss=68.124, grad_norm=3.956, loss_scale=1.000, learning_rate=1.146e-04, step_time=0.354 [2024-12-05 17:00:45,256] [INFO] [logging.py:129:log_dist] [Rank 0] step=183000, skipped=0, lr=[np.float64(0.00011451935396879135)], mom=[[0.9, 0.98]] [2024-12-05 17:00:45,257] [INFO] [timer.py:264:stop] epoch=0/micro_step=3000/global_step=3000, RunningAvgSamplesPerSec=45.54686632081202, CurrSamplesPerSec=47.82541019119195, MemAllocated=1.77GB, MaxMemAllocated=16.04GB [cnode7-012:0/16] 2024-12-05 17:00:45,259 (deepspeed_trainer:228) INFO: 13epoch:train:2901-3000batch: iter_time=1.122e-04, loss_ctc=83.591, loss_att=65.398, acc=0.670, loss=70.853, grad_norm=3.928, loss_scale=1.000, learning_rate=1.145e-04, step_time=0.354 [cnode7-012:0/16] 2024-12-05 17:01:20,574 (deepspeed_trainer:228) INFO: 13epoch:train:3001-3100batch: iter_time=1.116e-04, loss_ctc=88.549, loss_att=58.897, acc=0.667, loss=67.768, grad_norm=4.556, loss_scale=1.000, learning_rate=1.145e-04, step_time=0.353 [cnode7-012:0/16] 2024-12-05 17:01:56,257 (deepspeed_trainer:228) INFO: 13epoch:train:3101-3200batch: iter_time=1.080e-04, loss_ctc=95.210, loss_att=77.882, acc=0.637, loss=83.081, grad_norm=4.843, loss_scale=1.000, learning_rate=1.145e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-05 17:02:31,517 (deepspeed_trainer:228) INFO: 13epoch:train:3201-3300batch: iter_time=1.075e-04, loss_ctc=79.447, loss_att=64.801, acc=0.659, loss=69.217, grad_norm=3.883, loss_scale=1.000, learning_rate=1.144e-04, step_time=0.352 [cnode7-012:0/16] 2024-12-05 17:03:06,905 (deepspeed_trainer:228) INFO: 13epoch:train:3301-3400batch: iter_time=1.087e-04, loss_ctc=80.543, loss_att=60.776, acc=0.664, loss=66.722, grad_norm=3.909, loss_scale=1.000, learning_rate=1.144e-04, step_time=0.354 [cnode7-012:0/16] 2024-12-05 17:03:42,306 (deepspeed_trainer:228) INFO: 13epoch:train:3401-3500batch: iter_time=1.094e-04, loss_ctc=84.357, loss_att=65.238, acc=0.674, loss=70.996, grad_norm=3.763, loss_scale=1.000, learning_rate=1.144e-04, step_time=0.353 [cnode7-012:0/16] 2024-12-05 17:04:17,963 (deepspeed_trainer:228) INFO: 13epoch:train:3501-3600batch: iter_time=1.094e-04, loss_ctc=92.908, loss_att=72.998, acc=0.662, loss=78.976, grad_norm=4.045, loss_scale=1.000, learning_rate=1.143e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-05 17:04:53,352 (deepspeed_trainer:228) INFO: 13epoch:train:3601-3700batch: iter_time=1.097e-04, loss_ctc=78.504, loss_att=56.502, acc=0.666, loss=63.105, grad_norm=3.844, loss_scale=1.000, learning_rate=1.143e-04, step_time=0.353 [cnode7-012:0/16] 2024-12-05 17:05:13,374 (multiple_iter_factory:32) INFO: Building 2th iter-factory... [cnode7-012:0/16] 2024-12-05 17:05:39,268 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-05 17:05:54,709 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.6", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.6", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.6", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.6", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-05 17:05:54,709 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.6, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-05 17:05:54,712 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-05 17:06:21,336] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:06:21,949] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:06:22,972] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:06:23,319] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:06:23,509] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:06:23,564] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:06:23,779] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:06:23,935] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:06:24,011] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:06:24,381] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:06:24,362] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:06:24,877] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:06:24,939] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:06:25,294] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:06:25,628] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:06:27,410] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:07:09,201] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:07:10,673] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:07:11,136] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:07:11,404] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:07:11,411] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:07:11,561] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:07:11,797] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:07:11,802] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:07:12,206] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:07:12,487] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:07:12,709] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:07:13,487] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:07:13,646] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:07:15,872] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:07:16,353] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:07:16,832] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:07:56,110] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:07:57,297] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:07:57,861] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:07:58,528] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:07:58,741] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:07:59,361] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:07:59,395] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:07:59,440] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:07:59,946] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:08:00,898] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:08:00,974] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:08:01,453] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:08:03,277] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:08:04,839] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:08:05,695] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:08:07,284] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:08:42,455] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:08:43,472] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:08:45,304] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:08:45,725] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:08:46,385] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:08:46,805] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:08:47,302] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:08:47,549] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:08:48,469] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:08:48,668] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:08:48,904] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:08:51,215] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:08:51,372] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:08:54,707] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:08:55,425] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:08:56,842] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-05 17:09:41,118 (deepspeed_trainer:228) INFO: 13epoch:train:3701-3800batch: iter_time=2.375, loss_ctc=76.181, loss_att=58.647, acc=0.678, loss=63.905, grad_norm=3.950, loss_scale=1.000, learning_rate=1.143e-04, step_time=0.502 [cnode7-012:0/16] 2024-12-05 17:10:17,836 (deepspeed_trainer:228) INFO: 13epoch:train:3801-3900batch: iter_time=1.074e-04, loss_ctc=81.975, loss_att=70.381, acc=0.669, loss=73.853, grad_norm=4.157, loss_scale=1.000, learning_rate=1.143e-04, step_time=0.367 [2024-12-05 17:10:54,459] [INFO] [logging.py:129:log_dist] [Rank 0] step=184000, skipped=0, lr=[np.float64(0.00011420773779734023)], mom=[[0.9, 0.98]] [2024-12-05 17:10:54,460] [INFO] [timer.py:264:stop] epoch=0/micro_step=4000/global_step=4000, RunningAvgSamplesPerSec=45.20702071159699, CurrSamplesPerSec=43.76919491970642, MemAllocated=1.77GB, MaxMemAllocated=16.24GB [cnode7-012:0/16] 2024-12-05 17:10:54,462 (deepspeed_trainer:228) INFO: 13epoch:train:3901-4000batch: iter_time=1.095e-04, loss_ctc=86.520, loss_att=66.948, acc=0.651, loss=72.850, grad_norm=4.190, loss_scale=1.000, learning_rate=1.142e-04, step_time=0.366 [cnode7-012:0/16] 2024-12-05 17:11:30,990 (deepspeed_trainer:228) INFO: 13epoch:train:4001-4100batch: iter_time=1.100e-04, loss_ctc=86.519, loss_att=64.204, acc=0.657, loss=70.940, grad_norm=4.474, loss_scale=1.000, learning_rate=1.142e-04, step_time=0.365 [cnode7-012:0/16] 2024-12-05 17:12:07,279 (deepspeed_trainer:228) INFO: 13epoch:train:4101-4200batch: iter_time=1.117e-04, loss_ctc=71.947, loss_att=54.060, acc=0.677, loss=59.402, grad_norm=3.737, loss_scale=1.000, learning_rate=1.142e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-05 17:12:43,613 (deepspeed_trainer:228) INFO: 13epoch:train:4201-4300batch: iter_time=1.092e-04, loss_ctc=77.396, loss_att=60.880, acc=0.670, loss=65.844, grad_norm=3.940, loss_scale=1.000, learning_rate=1.141e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-05 17:13:20,194 (deepspeed_trainer:228) INFO: 13epoch:train:4301-4400batch: iter_time=1.071e-04, loss_ctc=95.996, loss_att=75.319, acc=0.658, loss=81.552, grad_norm=4.287, loss_scale=1.000, learning_rate=1.141e-04, step_time=0.365 [cnode7-012:0/16] 2024-12-05 17:13:56,353 (deepspeed_trainer:228) INFO: 13epoch:train:4401-4500batch: iter_time=1.103e-04, loss_ctc=84.882, loss_att=60.547, acc=0.676, loss=67.844, grad_norm=3.911, loss_scale=1.000, learning_rate=1.141e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 17:14:32,629 (deepspeed_trainer:228) INFO: 13epoch:train:4501-4600batch: iter_time=1.086e-04, loss_ctc=79.228, loss_att=70.648, acc=0.658, loss=73.224, grad_norm=3.795, loss_scale=1.000, learning_rate=1.140e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 17:15:09,035 (deepspeed_trainer:228) INFO: 13epoch:train:4601-4700batch: iter_time=1.110e-04, loss_ctc=100.916, loss_att=83.484, acc=0.659, loss=88.677, grad_norm=4.539, loss_scale=1.000, learning_rate=1.140e-04, step_time=0.364 [cnode7-012:0/16] 2024-12-05 17:15:44,919 (deepspeed_trainer:228) INFO: 13epoch:train:4701-4800batch: iter_time=1.080e-04, loss_ctc=81.474, loss_att=57.795, acc=0.679, loss=64.919, grad_norm=3.935, loss_scale=1.000, learning_rate=1.140e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 17:16:20,827 (deepspeed_trainer:228) INFO: 13epoch:train:4801-4900batch: iter_time=1.110e-04, loss_ctc=76.441, loss_att=64.814, acc=0.665, loss=68.265, grad_norm=3.582, loss_scale=1.000, learning_rate=1.139e-04, step_time=0.358 [2024-12-05 17:16:56,825] [INFO] [logging.py:129:log_dist] [Rank 0] step=185000, skipped=0, lr=[np.float64(0.0001138986516565222)], mom=[[0.9, 0.98]] [2024-12-05 17:16:56,826] [INFO] [timer.py:264:stop] epoch=0/micro_step=5000/global_step=5000, RunningAvgSamplesPerSec=45.244647149730504, CurrSamplesPerSec=46.19540834453468, MemAllocated=1.77GB, MaxMemAllocated=16.24GB [cnode7-012:0/16] 2024-12-05 17:16:56,828 (deepspeed_trainer:228) INFO: 13epoch:train:4901-5000batch: iter_time=1.121e-04, loss_ctc=98.032, loss_att=68.974, acc=0.667, loss=77.681, grad_norm=4.965, loss_scale=1.000, learning_rate=1.139e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 17:17:32,784 (deepspeed_trainer:228) INFO: 13epoch:train:5001-5100batch: iter_time=1.094e-04, loss_ctc=89.031, loss_att=71.443, acc=0.650, loss=76.710, grad_norm=4.817, loss_scale=1.000, learning_rate=1.139e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 17:18:08,320 (deepspeed_trainer:228) INFO: 13epoch:train:5101-5200batch: iter_time=1.102e-04, loss_ctc=75.267, loss_att=62.649, acc=0.669, loss=66.446, grad_norm=3.841, loss_scale=1.000, learning_rate=1.139e-04, step_time=0.355 [cnode7-012:0/16] 2024-12-05 17:18:44,114 (deepspeed_trainer:228) INFO: 13epoch:train:5201-5300batch: iter_time=1.098e-04, loss_ctc=83.638, loss_att=63.044, acc=0.677, loss=69.198, grad_norm=3.951, loss_scale=1.000, learning_rate=1.138e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-05 17:19:20,058 (deepspeed_trainer:228) INFO: 13epoch:train:5301-5400batch: iter_time=1.100e-04, loss_ctc=81.860, loss_att=68.896, acc=0.671, loss=72.808, grad_norm=3.985, loss_scale=1.000, learning_rate=1.138e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 17:19:56,036 (deepspeed_trainer:228) INFO: 13epoch:train:5401-5500batch: iter_time=1.094e-04, loss_ctc=96.181, loss_att=74.034, acc=0.670, loss=80.685, grad_norm=4.104, loss_scale=1.000, learning_rate=1.138e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 17:20:31,448 (deepspeed_trainer:228) INFO: 13epoch:train:5501-5600batch: iter_time=1.109e-04, loss_ctc=73.184, loss_att=52.914, acc=0.672, loss=58.981, grad_norm=3.704, loss_scale=1.000, learning_rate=1.137e-04, step_time=0.354 [cnode7-012:0/16] 2024-12-05 17:20:42,788 (multiple_iter_factory:32) INFO: Building 3th iter-factory... [cnode7-012:0/16] 2024-12-05 17:21:08,587 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-05 17:21:24,142 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.1", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.1", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.1", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.1", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-05 17:21:24,142 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.1, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-05 17:21:24,144 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-05 17:21:50,585] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:21:50,694] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:21:51,562] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:21:51,567] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:21:51,839] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:21:52,030] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:21:52,536] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:21:52,877] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:21:53,140] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:21:53,643] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:21:53,684] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:21:53,843] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:21:53,984] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:21:54,229] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:21:54,340] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:21:54,657] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:22:38,087] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:22:38,556] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:22:38,858] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:22:39,083] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:22:39,453] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:22:39,778] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:22:40,774] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:22:41,268] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:22:41,312] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:22:42,026] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:22:42,382] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:22:42,574] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:22:42,685] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:22:43,161] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:22:43,379] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:22:45,922] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:23:24,438] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:23:25,610] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:23:25,887] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:23:26,185] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:23:27,333] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:23:27,835] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:23:28,094] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:23:28,892] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:23:29,768] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:23:30,160] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:23:30,739] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:23:31,138] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:23:31,263] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:23:31,694] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:23:31,761] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:23:36,238] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:24:11,359] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:24:12,291] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:24:12,339] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:24:12,960] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:24:14,764] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:24:15,259] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:24:15,715] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:24:16,480] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:24:17,257] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:24:18,866] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:24:19,106] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:24:19,681] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:24:19,618] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:24:19,931] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:24:20,338] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:24:25,764] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-05 17:25:17,935 (deepspeed_trainer:228) INFO: 13epoch:train:5601-5700batch: iter_time=2.378, loss_ctc=79.252, loss_att=65.547, acc=0.667, loss=69.683, grad_norm=3.807, loss_scale=1.000, learning_rate=1.137e-04, step_time=0.486 [cnode7-012:0/16] 2024-12-05 17:25:53,913 (deepspeed_trainer:228) INFO: 13epoch:train:5701-5800batch: iter_time=1.082e-04, loss_ctc=80.636, loss_att=63.010, acc=0.664, loss=68.296, grad_norm=3.894, loss_scale=1.000, learning_rate=1.137e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 17:26:29,955 (deepspeed_trainer:228) INFO: 13epoch:train:5801-5900batch: iter_time=1.106e-04, loss_ctc=90.863, loss_att=70.276, acc=0.650, loss=76.463, grad_norm=4.723, loss_scale=1.000, learning_rate=1.136e-04, step_time=0.360 [2024-12-05 17:27:05,860] [INFO] [logging.py:129:log_dist] [Rank 0] step=186000, skipped=0, lr=[np.float64(0.00011359206149481937)], mom=[[0.9, 0.98]] [2024-12-05 17:27:05,861] [INFO] [timer.py:264:stop] epoch=0/micro_step=6000/global_step=6000, RunningAvgSamplesPerSec=45.08459088843663, CurrSamplesPerSec=42.23610061398322, MemAllocated=1.77GB, MaxMemAllocated=16.24GB [cnode7-012:0/16] 2024-12-05 17:27:05,863 (deepspeed_trainer:228) INFO: 13epoch:train:5901-6000batch: iter_time=1.111e-04, loss_ctc=79.131, loss_att=61.936, acc=0.659, loss=67.070, grad_norm=4.499, loss_scale=1.000, learning_rate=1.136e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 17:27:41,543 (deepspeed_trainer:228) INFO: 13epoch:train:6001-6100batch: iter_time=1.121e-04, loss_ctc=73.451, loss_att=52.265, acc=0.673, loss=58.612, grad_norm=3.884, loss_scale=1.000, learning_rate=1.136e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-05 17:28:17,400 (deepspeed_trainer:228) INFO: 13epoch:train:6101-6200batch: iter_time=1.128e-04, loss_ctc=78.661, loss_att=63.765, acc=0.657, loss=68.271, grad_norm=4.119, loss_scale=1.000, learning_rate=1.135e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 17:28:53,547 (deepspeed_trainer:228) INFO: 13epoch:train:6201-6300batch: iter_time=1.094e-04, loss_ctc=93.843, loss_att=69.139, acc=0.665, loss=76.542, grad_norm=4.659, loss_scale=1.000, learning_rate=1.135e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 17:29:29,625 (deepspeed_trainer:228) INFO: 13epoch:train:6301-6400batch: iter_time=1.120e-04, loss_ctc=82.255, loss_att=63.859, acc=0.661, loss=69.362, grad_norm=4.395, loss_scale=1.000, learning_rate=1.135e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 17:30:05,788 (deepspeed_trainer:228) INFO: 13epoch:train:6401-6500batch: iter_time=1.118e-04, loss_ctc=89.968, loss_att=76.932, acc=0.656, loss=80.869, grad_norm=4.403, loss_scale=1.000, learning_rate=1.135e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 17:30:41,788 (deepspeed_trainer:228) INFO: 13epoch:train:6501-6600batch: iter_time=1.106e-04, loss_ctc=86.872, loss_att=66.724, acc=0.663, loss=72.767, grad_norm=4.396, loss_scale=1.000, learning_rate=1.134e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 17:31:17,632 (deepspeed_trainer:228) INFO: 13epoch:train:6601-6700batch: iter_time=1.118e-04, loss_ctc=86.793, loss_att=63.375, acc=0.673, loss=70.404, grad_norm=4.234, loss_scale=1.000, learning_rate=1.134e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 17:31:53,278 (deepspeed_trainer:228) INFO: 13epoch:train:6701-6800batch: iter_time=1.123e-04, loss_ctc=74.623, loss_att=62.078, acc=0.660, loss=65.873, grad_norm=3.846, loss_scale=1.000, learning_rate=1.134e-04, step_time=0.356 [cnode7-012:0/16] 2024-12-05 17:32:29,197 (deepspeed_trainer:228) INFO: 13epoch:train:6801-6900batch: iter_time=1.130e-04, loss_ctc=96.457, loss_att=70.537, acc=0.658, loss=78.321, grad_norm=5.077, loss_scale=1.000, learning_rate=1.133e-04, step_time=0.359 [2024-12-05 17:33:04,969] [INFO] [logging.py:129:log_dist] [Rank 0] step=187000, skipped=0, lr=[np.float64(0.00011328793389889495)], mom=[[0.9, 0.98]] [2024-12-05 17:33:04,970] [INFO] [timer.py:264:stop] epoch=0/micro_step=7000/global_step=7000, RunningAvgSamplesPerSec=45.189674012035574, CurrSamplesPerSec=42.69779846094874, MemAllocated=1.77GB, MaxMemAllocated=16.24GB [cnode7-012:0/16] 2024-12-05 17:33:04,972 (deepspeed_trainer:228) INFO: 13epoch:train:6901-7000batch: iter_time=1.111e-04, loss_ctc=87.777, loss_att=66.970, acc=0.649, loss=73.226, grad_norm=4.728, loss_scale=1.000, learning_rate=1.133e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-05 17:33:41,002 (deepspeed_trainer:228) INFO: 13epoch:train:7001-7100batch: iter_time=1.126e-04, loss_ctc=76.713, loss_att=61.158, acc=0.665, loss=65.841, grad_norm=3.823, loss_scale=1.000, learning_rate=1.133e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 17:34:17,183 (deepspeed_trainer:228) INFO: 13epoch:train:7101-7200batch: iter_time=1.100e-04, loss_ctc=81.885, loss_att=66.784, acc=0.672, loss=71.349, grad_norm=3.834, loss_scale=1.000, learning_rate=1.132e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 17:34:53,406 (deepspeed_trainer:228) INFO: 13epoch:train:7201-7300batch: iter_time=1.123e-04, loss_ctc=85.676, loss_att=63.348, acc=0.675, loss=70.049, grad_norm=3.893, loss_scale=1.000, learning_rate=1.132e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 17:35:29,391 (deepspeed_trainer:228) INFO: 13epoch:train:7301-7400batch: iter_time=1.094e-04, loss_ctc=91.658, loss_att=68.312, acc=0.668, loss=75.295, grad_norm=4.128, loss_scale=1.000, learning_rate=1.132e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 17:36:05,010 (deepspeed_trainer:228) INFO: 13epoch:train:7401-7500batch: iter_time=1.093e-04, loss_ctc=72.103, loss_att=53.367, acc=0.674, loss=59.009, grad_norm=3.756, loss_scale=1.000, learning_rate=1.132e-04, step_time=0.356 [cnode7-012:0/16] 2024-12-05 17:36:07,486 (multiple_iter_factory:32) INFO: Building 4th iter-factory... [cnode7-012:0/16] 2024-12-05 17:36:33,803 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-05 17:36:49,129 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.3", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.3", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.3", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.3", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-05 17:36:49,129 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.3, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-05 17:36:49,132 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-05 17:37:15,481] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:37:15,841] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:37:15,834] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:37:16,177] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:37:16,210] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:37:16,621] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:37:16,679] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:37:17,552] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:37:17,850] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:37:18,012] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:37:18,300] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:37:18,390] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:37:18,607] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:37:19,137] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:37:19,190] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:37:19,723] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:38:02,950] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:38:02,955] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:38:03,336] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:38:03,589] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:38:03,590] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:38:04,195] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:38:04,779] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:38:05,487] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:38:05,832] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:38:06,019] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:38:06,136] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:38:06,290] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:38:07,291] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:38:07,678] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:38:07,723] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:38:10,071] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:38:50,764] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:38:50,800] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:38:51,884] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:38:52,269] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:38:53,023] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:38:53,073] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:38:53,356] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:38:53,776] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:38:54,152] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:38:54,562] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:38:54,749] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:38:54,874] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:38:55,334] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:38:55,596] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:38:55,937] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:39:00,454] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:39:37,970] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:39:38,148] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:39:38,689] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:39:40,160] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:39:40,732] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:39:40,926] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:39:41,560] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:39:41,821] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:39:42,233] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:39:42,866] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:39:43,202] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:39:43,366] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:39:43,449] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:39:43,823] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:39:44,264] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:39:49,244] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-05 17:40:50,810 (deepspeed_trainer:228) INFO: 13epoch:train:7501-7600batch: iter_time=2.436, loss_ctc=81.762, loss_att=66.369, acc=0.679, loss=70.984, grad_norm=3.691, loss_scale=1.000, learning_rate=1.131e-04, step_time=0.422 [cnode7-012:0/16] 2024-12-05 17:41:26,727 (deepspeed_trainer:228) INFO: 13epoch:train:7601-7700batch: iter_time=1.081e-04, loss_ctc=76.330, loss_att=55.546, acc=0.667, loss=61.786, grad_norm=3.894, loss_scale=1.000, learning_rate=1.131e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 17:42:02,832 (deepspeed_trainer:228) INFO: 13epoch:train:7701-7800batch: iter_time=1.058e-04, loss_ctc=89.694, loss_att=72.599, acc=0.649, loss=77.685, grad_norm=4.744, loss_scale=1.000, learning_rate=1.131e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 17:42:38,718 (deepspeed_trainer:228) INFO: 13epoch:train:7801-7900batch: iter_time=1.098e-04, loss_ctc=78.918, loss_att=59.829, acc=0.660, loss=65.556, grad_norm=4.229, loss_scale=1.000, learning_rate=1.130e-04, step_time=0.358 [2024-12-05 17:43:14,539] [INFO] [logging.py:129:log_dist] [Rank 0] step=188000, skipped=0, lr=[np.float64(0.0001129862360782973)], mom=[[0.9, 0.98]] [2024-12-05 17:43:14,541] [INFO] [timer.py:264:stop] epoch=0/micro_step=8000/global_step=8000, RunningAvgSamplesPerSec=45.1562192375989, CurrSamplesPerSec=44.400520147171505, MemAllocated=1.77GB, MaxMemAllocated=16.24GB [cnode7-012:0/16] 2024-12-05 17:43:14,544 (deepspeed_trainer:228) INFO: 13epoch:train:7901-8000batch: iter_time=1.133e-04, loss_ctc=74.826, loss_att=55.779, acc=0.676, loss=61.474, grad_norm=3.713, loss_scale=1.000, learning_rate=1.130e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 17:43:50,485 (deepspeed_trainer:228) INFO: 13epoch:train:8001-8100batch: iter_time=1.125e-04, loss_ctc=83.697, loss_att=66.177, acc=0.665, loss=71.419, grad_norm=4.261, loss_scale=1.000, learning_rate=1.130e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 17:44:26,803 (deepspeed_trainer:228) INFO: 13epoch:train:8101-8200batch: iter_time=1.145e-04, loss_ctc=86.581, loss_att=60.871, acc=0.667, loss=68.608, grad_norm=4.404, loss_scale=1.000, learning_rate=1.129e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-05 17:45:02,971 (deepspeed_trainer:228) INFO: 13epoch:train:8201-8300batch: iter_time=1.109e-04, loss_ctc=84.506, loss_att=71.488, acc=0.660, loss=75.385, grad_norm=4.307, loss_scale=1.000, learning_rate=1.129e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 17:45:39,468 (deepspeed_trainer:228) INFO: 13epoch:train:8301-8400batch: iter_time=1.135e-04, loss_ctc=95.669, loss_att=72.676, acc=0.659, loss=79.594, grad_norm=4.719, loss_scale=1.000, learning_rate=1.129e-04, step_time=0.364 [cnode7-012:0/16] 2024-12-05 17:46:15,850 (deepspeed_trainer:228) INFO: 13epoch:train:8401-8500batch: iter_time=1.132e-04, loss_ctc=77.981, loss_att=64.650, acc=0.662, loss=68.642, grad_norm=4.211, loss_scale=1.000, learning_rate=1.129e-04, step_time=0.364 [cnode7-012:0/16] 2024-12-05 17:46:52,369 (deepspeed_trainer:228) INFO: 13epoch:train:8501-8600batch: iter_time=1.099e-04, loss_ctc=86.034, loss_att=61.269, acc=0.684, loss=68.743, grad_norm=3.997, loss_scale=1.000, learning_rate=1.128e-04, step_time=0.365 [cnode7-012:0/16] 2024-12-05 17:47:28,617 (deepspeed_trainer:228) INFO: 13epoch:train:8601-8700batch: iter_time=1.112e-04, loss_ctc=74.612, loss_att=60.034, acc=0.667, loss=64.371, grad_norm=4.343, loss_scale=1.000, learning_rate=1.128e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 17:48:05,159 (deepspeed_trainer:228) INFO: 13epoch:train:8701-8800batch: iter_time=1.115e-04, loss_ctc=100.246, loss_att=74.673, acc=0.655, loss=82.302, grad_norm=5.021, loss_scale=1.000, learning_rate=1.128e-04, step_time=0.365 [cnode7-012:0/16] 2024-12-05 17:48:41,378 (deepspeed_trainer:228) INFO: 13epoch:train:8801-8900batch: iter_time=1.125e-04, loss_ctc=85.347, loss_att=66.429, acc=0.652, loss=72.096, grad_norm=4.879, loss_scale=1.000, learning_rate=1.127e-04, step_time=0.362 [2024-12-05 17:49:17,707] [INFO] [logging.py:129:log_dist] [Rank 0] step=189000, skipped=0, lr=[np.float64(0.00011268693585060964)], mom=[[0.9, 0.98]] [2024-12-05 17:49:17,707] [INFO] [timer.py:264:stop] epoch=0/micro_step=9000/global_step=9000, RunningAvgSamplesPerSec=45.16876062732037, CurrSamplesPerSec=43.37074546017377, MemAllocated=1.77GB, MaxMemAllocated=16.24GB [cnode7-012:0/16] 2024-12-05 17:49:17,709 (deepspeed_trainer:228) INFO: 13epoch:train:8901-9000batch: iter_time=1.109e-04, loss_ctc=74.311, loss_att=57.622, acc=0.667, loss=62.601, grad_norm=3.879, loss_scale=1.000, learning_rate=1.127e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-05 17:49:54,227 (deepspeed_trainer:228) INFO: 13epoch:train:9001-9100batch: iter_time=1.092e-04, loss_ctc=83.063, loss_att=67.489, acc=0.677, loss=72.170, grad_norm=4.226, loss_scale=1.000, learning_rate=1.127e-04, step_time=0.365 [cnode7-012:0/16] 2024-12-05 17:50:30,804 (deepspeed_trainer:228) INFO: 13epoch:train:9101-9200batch: iter_time=1.093e-04, loss_ctc=90.386, loss_att=65.425, acc=0.676, loss=72.939, grad_norm=4.103, loss_scale=1.000, learning_rate=1.126e-04, step_time=0.365 [cnode7-012:0/16] 2024-12-05 17:51:07,189 (deepspeed_trainer:228) INFO: 13epoch:train:9201-9300batch: iter_time=1.106e-04, loss_ctc=84.925, loss_att=64.745, acc=0.670, loss=70.798, grad_norm=4.039, loss_scale=1.000, learning_rate=1.126e-04, step_time=0.364 [cnode7-012:0/16] 2024-12-05 17:51:36,702 (multiple_iter_factory:32) INFO: Building 5th iter-factory... [cnode7-012:0/16] 2024-12-05 17:52:02,704 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-05 17:52:17,798 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.0", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.0", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.0", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.0", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-05 17:52:17,798 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.0, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-05 17:52:17,803 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-05 17:52:45,458] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:52:45,782] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:52:46,559] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:52:46,613] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:52:46,807] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:52:46,959] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:52:46,990] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:52:47,328] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:52:47,409] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:52:47,824] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:52:48,131] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:52:48,699] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:52:49,213] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:52:49,251] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:52:49,485] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:52:50,283] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:53:33,740] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:53:33,979] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:53:34,334] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:53:34,645] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:53:35,006] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:53:35,199] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:53:35,415] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:53:35,606] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:53:35,837] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:53:36,441] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:53:36,734] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:53:36,934] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:53:37,678] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:53:38,027] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:53:38,164] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:53:38,218] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:54:21,013] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:54:21,269] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:54:21,819] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:54:22,752] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:54:23,151] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:54:23,255] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:54:23,869] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:54:24,145] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:54:24,275] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:54:24,898] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:54:26,060] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:54:26,083] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:54:26,299] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:54:26,660] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:54:26,918] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:54:29,011] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:55:08,671] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:55:09,604] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:55:10,028] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:55:10,765] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:55:11,687] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:55:11,747] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:55:11,834] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:55:11,960] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:55:12,214] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:55:13,069] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:55:14,168] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:55:15,324] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:55:15,390] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:55:15,494] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:55:15,909] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 17:55:19,761] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-05 17:55:54,350 (deepspeed_trainer:228) INFO: 13epoch:train:9301-9400batch: iter_time=2.423, loss_ctc=71.274, loss_att=53.885, acc=0.677, loss=59.108, grad_norm=4.273, loss_scale=1.000, learning_rate=1.126e-04, step_time=0.449 [cnode7-012:0/16] 2024-12-05 17:56:30,529 (deepspeed_trainer:228) INFO: 13epoch:train:9401-9500batch: iter_time=1.088e-04, loss_ctc=80.645, loss_att=65.046, acc=0.683, loss=69.701, grad_norm=4.124, loss_scale=1.000, learning_rate=1.126e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 17:57:06,689 (deepspeed_trainer:228) INFO: 13epoch:train:9501-9600batch: iter_time=1.111e-04, loss_ctc=80.563, loss_att=64.644, acc=0.664, loss=69.437, grad_norm=4.050, loss_scale=1.000, learning_rate=1.125e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 17:57:42,792 (deepspeed_trainer:228) INFO: 13epoch:train:9601-9700batch: iter_time=1.109e-04, loss_ctc=85.275, loss_att=69.254, acc=0.649, loss=74.069, grad_norm=4.982, loss_scale=1.000, learning_rate=1.125e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 17:58:18,586 (deepspeed_trainer:228) INFO: 13epoch:train:9701-9800batch: iter_time=1.098e-04, loss_ctc=74.832, loss_att=55.494, acc=0.673, loss=61.262, grad_norm=4.066, loss_scale=1.000, learning_rate=1.125e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-05 17:58:54,478 (deepspeed_trainer:228) INFO: 13epoch:train:9801-9900batch: iter_time=1.103e-04, loss_ctc=75.642, loss_att=60.481, acc=0.672, loss=65.042, grad_norm=3.894, loss_scale=1.000, learning_rate=1.124e-04, step_time=0.359 [2024-12-05 17:59:30,695] [INFO] [logging.py:129:log_dist] [Rank 0] step=190000, skipped=0, lr=[np.float64(0.00011239000162703026)], mom=[[0.9, 0.98]] [2024-12-05 17:59:30,696] [INFO] [timer.py:264:stop] epoch=0/micro_step=10000/global_step=10000, RunningAvgSamplesPerSec=45.08479637667568, CurrSamplesPerSec=46.43758289121401, MemAllocated=1.77GB, MaxMemAllocated=17.74GB [cnode7-012:0/16] 2024-12-05 17:59:30,698 (deepspeed_trainer:228) INFO: 13epoch:train:9901-10000batch: iter_time=1.119e-04, loss_ctc=86.518, loss_att=69.228, acc=0.666, loss=74.397, grad_norm=4.323, loss_scale=1.000, learning_rate=1.124e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 18:00:07,061 (deepspeed_trainer:228) INFO: 13epoch:train:10001-10100batch: iter_time=1.126e-04, loss_ctc=87.247, loss_att=62.545, acc=0.677, loss=69.961, grad_norm=4.506, loss_scale=1.000, learning_rate=1.124e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-05 18:00:42,974 (deepspeed_trainer:228) INFO: 13epoch:train:10101-10200batch: iter_time=1.123e-04, loss_ctc=79.533, loss_att=67.957, acc=0.666, loss=71.439, grad_norm=3.860, loss_scale=1.000, learning_rate=1.123e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 18:01:19,384 (deepspeed_trainer:228) INFO: 13epoch:train:10201-10300batch: iter_time=1.116e-04, loss_ctc=98.877, loss_att=79.254, acc=0.668, loss=85.133, grad_norm=5.055, loss_scale=1.000, learning_rate=1.123e-04, step_time=0.364 [cnode7-012:0/16] 2024-12-05 18:01:55,255 (deepspeed_trainer:228) INFO: 13epoch:train:10301-10400batch: iter_time=1.094e-04, loss_ctc=78.486, loss_att=61.465, acc=0.678, loss=66.574, grad_norm=4.090, loss_scale=1.000, learning_rate=1.123e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 18:02:31,149 (deepspeed_trainer:228) INFO: 13epoch:train:10401-10500batch: iter_time=1.130e-04, loss_ctc=81.690, loss_att=66.053, acc=0.680, loss=70.779, grad_norm=4.091, loss_scale=1.000, learning_rate=1.123e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 18:03:06,991 (deepspeed_trainer:228) INFO: 13epoch:train:10501-10600batch: iter_time=1.111e-04, loss_ctc=86.604, loss_att=57.910, acc=0.678, loss=66.508, grad_norm=4.607, loss_scale=1.000, learning_rate=1.122e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 18:03:43,128 (deepspeed_trainer:228) INFO: 13epoch:train:10601-10700batch: iter_time=1.093e-04, loss_ctc=92.574, loss_att=77.471, acc=0.649, loss=82.014, grad_norm=4.830, loss_scale=1.000, learning_rate=1.122e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 18:04:18,815 (deepspeed_trainer:228) INFO: 13epoch:train:10701-10800batch: iter_time=1.121e-04, loss_ctc=77.770, loss_att=64.535, acc=0.669, loss=68.465, grad_norm=4.123, loss_scale=1.000, learning_rate=1.122e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-05 18:04:54,640 (deepspeed_trainer:228) INFO: 13epoch:train:10801-10900batch: iter_time=1.114e-04, loss_ctc=78.920, loss_att=60.640, acc=0.676, loss=66.138, grad_norm=4.117, loss_scale=1.000, learning_rate=1.121e-04, step_time=0.358 [2024-12-05 18:05:30,664] [INFO] [logging.py:129:log_dist] [Rank 0] step=191000, skipped=0, lr=[np.float64(0.00011209540239836847)], mom=[[0.9, 0.98]] [2024-12-05 18:05:30,664] [INFO] [timer.py:264:stop] epoch=0/micro_step=11000/global_step=11000, RunningAvgSamplesPerSec=45.14359981467134, CurrSamplesPerSec=43.67928632708927, MemAllocated=1.77GB, MaxMemAllocated=17.74GB [cnode7-012:0/16] 2024-12-05 18:05:30,666 (deepspeed_trainer:228) INFO: 13epoch:train:10901-11000batch: iter_time=1.094e-04, loss_ctc=82.979, loss_att=63.974, acc=0.684, loss=69.656, grad_norm=3.835, loss_scale=1.000, learning_rate=1.121e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 18:06:06,689 (deepspeed_trainer:228) INFO: 13epoch:train:11001-11100batch: iter_time=1.106e-04, loss_ctc=90.632, loss_att=76.448, acc=0.666, loss=80.705, grad_norm=4.321, loss_scale=1.000, learning_rate=1.121e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 18:06:42,525 (deepspeed_trainer:228) INFO: 13epoch:train:11101-11200batch: iter_time=1.144e-04, loss_ctc=77.193, loss_att=55.015, acc=0.673, loss=61.683, grad_norm=4.050, loss_scale=1.000, learning_rate=1.121e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 18:07:02,790 (multiple_iter_factory:32) INFO: Building 6th iter-factory... [cnode7-012:0/16] 2024-12-05 18:07:29,006 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-05 18:07:45,599 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.2", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.2", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.2", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.2", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-05 18:07:45,599 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.2, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-05 18:07:45,602 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-05 18:08:11,747] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:08:11,841] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:08:12,096] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:08:12,436] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:08:12,528] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:08:12,794] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:08:13,331] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:08:13,702] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:08:13,781] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:08:13,894] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:08:13,937] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:08:14,425] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:08:14,435] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:08:14,676] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:08:15,619] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:08:16,134] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:08:59,013] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:08:59,140] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:08:59,472] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:09:00,060] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:09:00,576] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:09:02,144] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:09:02,264] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:09:02,389] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:09:02,918] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:09:02,942] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:09:03,138] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:09:03,206] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:09:03,228] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:09:03,319] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:09:03,667] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:09:04,045] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:09:46,215] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:09:46,293] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:09:46,630] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:09:46,680] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:09:48,459] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:09:49,665] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:09:50,273] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:09:50,817] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:09:50,988] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:09:51,130] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:09:51,392] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:09:51,942] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:09:52,006] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:09:52,091] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:09:52,232] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:09:52,351] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:10:33,182] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:10:33,483] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:10:33,801] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:10:34,966] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:10:37,048] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:10:38,275] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:10:38,835] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:10:39,049] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:10:39,440] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:10:40,021] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:10:40,064] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:10:40,313] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:10:40,327] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:10:40,464] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:10:40,545] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:10:40,857] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-05 18:11:25,814 (deepspeed_trainer:228) INFO: 13epoch:train:11201-11300batch: iter_time=2.467, loss_ctc=75.497, loss_att=57.263, acc=0.684, loss=62.754, grad_norm=4.102, loss_scale=1.000, learning_rate=1.120e-04, step_time=0.366 [cnode7-012:0/16] 2024-12-05 18:12:03,069 (deepspeed_trainer:228) INFO: 13epoch:train:11301-11400batch: iter_time=1.157e-04, loss_ctc=80.767, loss_att=68.955, acc=0.675, loss=72.519, grad_norm=3.731, loss_scale=1.000, learning_rate=1.120e-04, step_time=0.372 [cnode7-012:0/16] 2024-12-05 18:12:40,503 (deepspeed_trainer:228) INFO: 13epoch:train:11401-11500batch: iter_time=1.161e-04, loss_ctc=82.963, loss_att=64.772, acc=0.657, loss=70.271, grad_norm=4.507, loss_scale=1.000, learning_rate=1.120e-04, step_time=0.374 [cnode7-012:0/16] 2024-12-05 18:13:17,311 (deepspeed_trainer:228) INFO: 13epoch:train:11501-11600batch: iter_time=1.112e-04, loss_ctc=85.469, loss_att=63.328, acc=0.663, loss=69.995, grad_norm=4.949, loss_scale=1.000, learning_rate=1.119e-04, step_time=0.368 [cnode7-012:0/16] 2024-12-05 18:13:53,086 (deepspeed_trainer:228) INFO: 13epoch:train:11601-11700batch: iter_time=1.140e-04, loss_ctc=70.334, loss_att=53.092, acc=0.681, loss=58.278, grad_norm=3.784, loss_scale=1.000, learning_rate=1.119e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 18:14:28,764 (deepspeed_trainer:228) INFO: 13epoch:train:11701-11800batch: iter_time=1.137e-04, loss_ctc=75.668, loss_att=58.942, acc=0.678, loss=63.939, grad_norm=4.383, loss_scale=1.000, learning_rate=1.119e-04, step_time=0.356 [cnode7-012:0/16] 2024-12-05 18:15:04,726 (deepspeed_trainer:228) INFO: 13epoch:train:11801-11900batch: iter_time=1.130e-04, loss_ctc=94.383, loss_att=74.532, acc=0.662, loss=80.466, grad_norm=4.633, loss_scale=1.000, learning_rate=1.118e-04, step_time=0.361 [2024-12-05 18:15:40,524] [INFO] [logging.py:129:log_dist] [Rank 0] step=192000, skipped=0, lr=[np.float64(0.00011180310772144223)], mom=[[0.9, 0.98]] [2024-12-05 18:15:40,525] [INFO] [timer.py:264:stop] epoch=0/micro_step=12000/global_step=12000, RunningAvgSamplesPerSec=45.15803204052768, CurrSamplesPerSec=43.72200092946854, MemAllocated=1.77GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 18:15:40,527 (deepspeed_trainer:228) INFO: 13epoch:train:11901-12000batch: iter_time=1.119e-04, loss_ctc=83.068, loss_att=59.646, acc=0.679, loss=66.669, grad_norm=4.341, loss_scale=1.000, learning_rate=1.118e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-05 18:16:16,643 (deepspeed_trainer:228) INFO: 13epoch:train:12001-12100batch: iter_time=1.105e-04, loss_ctc=78.183, loss_att=69.111, acc=0.664, loss=71.862, grad_norm=3.961, loss_scale=1.000, learning_rate=1.118e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 18:16:53,393 (deepspeed_trainer:228) INFO: 13epoch:train:12101-12200batch: iter_time=1.127e-04, loss_ctc=99.957, loss_att=81.918, acc=0.666, loss=87.323, grad_norm=4.665, loss_scale=1.000, learning_rate=1.118e-04, step_time=0.367 [cnode7-012:0/16] 2024-12-05 18:17:29,382 (deepspeed_trainer:228) INFO: 13epoch:train:12201-12300batch: iter_time=1.131e-04, loss_ctc=79.591, loss_att=56.935, acc=0.684, loss=63.739, grad_norm=4.326, loss_scale=1.000, learning_rate=1.117e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 18:18:05,344 (deepspeed_trainer:228) INFO: 13epoch:train:12301-12400batch: iter_time=1.126e-04, loss_ctc=75.335, loss_att=63.880, acc=0.669, loss=67.332, grad_norm=3.762, loss_scale=1.000, learning_rate=1.117e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 18:18:41,696 (deepspeed_trainer:228) INFO: 13epoch:train:12401-12500batch: iter_time=1.116e-04, loss_ctc=95.966, loss_att=67.426, acc=0.673, loss=75.975, grad_norm=5.190, loss_scale=1.000, learning_rate=1.117e-04, step_time=0.364 [cnode7-012:0/16] 2024-12-05 18:19:17,965 (deepspeed_trainer:228) INFO: 13epoch:train:12501-12600batch: iter_time=1.137e-04, loss_ctc=87.254, loss_att=70.040, acc=0.656, loss=75.185, grad_norm=4.879, loss_scale=1.000, learning_rate=1.116e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 18:19:53,712 (deepspeed_trainer:228) INFO: 13epoch:train:12601-12700batch: iter_time=1.142e-04, loss_ctc=74.366, loss_att=61.547, acc=0.674, loss=65.401, grad_norm=3.889, loss_scale=1.000, learning_rate=1.116e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-05 18:20:30,056 (deepspeed_trainer:228) INFO: 13epoch:train:12701-12800batch: iter_time=1.158e-04, loss_ctc=82.074, loss_att=61.808, acc=0.683, loss=67.910, grad_norm=3.902, loss_scale=1.000, learning_rate=1.116e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-05 18:21:06,424 (deepspeed_trainer:228) INFO: 13epoch:train:12801-12900batch: iter_time=1.157e-04, loss_ctc=80.459, loss_att=67.460, acc=0.675, loss=71.362, grad_norm=3.929, loss_scale=1.000, learning_rate=1.116e-04, step_time=0.363 [2024-12-05 18:21:42,863] [INFO] [logging.py:129:log_dist] [Rank 0] step=193000, skipped=0, lr=[np.float64(0.00011151308770586395)], mom=[[0.9, 0.98]] [2024-12-05 18:21:42,864] [INFO] [timer.py:264:stop] epoch=0/micro_step=13000/global_step=13000, RunningAvgSamplesPerSec=45.1783353921134, CurrSamplesPerSec=48.16322425792337, MemAllocated=1.77GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 18:21:42,865 (deepspeed_trainer:228) INFO: 13epoch:train:12901-13000batch: iter_time=1.126e-04, loss_ctc=94.560, loss_att=73.100, acc=0.675, loss=79.544, grad_norm=4.372, loss_scale=1.000, learning_rate=1.115e-04, step_time=0.364 [cnode7-012:0/16] 2024-12-05 18:22:18,319 (deepspeed_trainer:228) INFO: 13epoch:train:13001-13100batch: iter_time=1.130e-04, loss_ctc=71.984, loss_att=51.708, acc=0.679, loss=57.795, grad_norm=3.712, loss_scale=1.000, learning_rate=1.115e-04, step_time=0.354 [cnode7-012:0/16] 2024-12-05 18:22:29,675 (multiple_iter_factory:32) INFO: Building 7th iter-factory... [cnode7-012:0/16] 2024-12-05 18:22:56,171 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-05 18:23:11,524 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.5", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.5", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.5", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.5", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-05 18:23:11,524 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.5, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-05 18:23:11,526 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-05 18:23:38,347] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:23:38,923] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:23:38,959] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:23:39,526] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:23:39,950] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:23:40,035] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:23:40,462] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:23:40,693] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:23:40,740] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:23:41,074] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:23:41,583] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:23:41,505] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:23:41,878] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:23:42,105] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:23:42,182] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:23:42,705] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:24:26,823] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:24:26,795] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:24:26,876] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:24:27,662] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:24:28,281] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:24:28,975] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:24:28,837] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:24:29,158] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:24:29,630] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:24:29,444] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:24:29,783] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:24:30,129] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:24:30,305] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:24:30,362] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:24:30,675] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:24:33,682] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:25:13,951] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:25:14,139] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:25:14,496] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:25:14,844] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:25:16,293] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:25:17,315] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:25:17,310] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:25:17,815] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:25:17,932] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:25:18,244] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:25:18,625] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:25:18,688] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:25:18,911] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:25:19,177] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:25:21,759] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:25:24,049] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:26:01,037] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:26:01,310] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:26:01,881] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:26:02,464] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:26:04,374] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:26:05,273] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:26:05,945] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:26:06,542] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:26:06,833] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:26:06,751] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:26:07,548] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:26:07,983] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:26:08,327] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:26:10,524] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:26:11,594] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:26:14,444] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-05 18:27:06,664 (deepspeed_trainer:228) INFO: 13epoch:train:13101-13200batch: iter_time=2.397, loss_ctc=78.627, loss_att=64.852, acc=0.672, loss=69.011, grad_norm=4.229, loss_scale=1.000, learning_rate=1.115e-04, step_time=0.486 [cnode7-012:0/16] 2024-12-05 18:27:43,057 (deepspeed_trainer:228) INFO: 13epoch:train:13201-13300batch: iter_time=1.110e-04, loss_ctc=78.626, loss_att=62.139, acc=0.668, loss=67.121, grad_norm=4.428, loss_scale=1.000, learning_rate=1.114e-04, step_time=0.364 [cnode7-012:0/16] 2024-12-05 18:28:18,997 (deepspeed_trainer:228) INFO: 13epoch:train:13301-13400batch: iter_time=1.142e-04, loss_ctc=88.217, loss_att=69.025, acc=0.654, loss=74.782, grad_norm=4.889, loss_scale=1.000, learning_rate=1.114e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 18:28:54,774 (deepspeed_trainer:228) INFO: 13epoch:train:13401-13500batch: iter_time=1.220e-04, loss_ctc=78.378, loss_att=61.178, acc=0.665, loss=66.318, grad_norm=4.192, loss_scale=1.000, learning_rate=1.114e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-05 18:29:30,352 (deepspeed_trainer:228) INFO: 13epoch:train:13501-13600batch: iter_time=1.132e-04, loss_ctc=72.296, loss_att=51.153, acc=0.680, loss=57.506, grad_norm=3.932, loss_scale=1.000, learning_rate=1.114e-04, step_time=0.356 [cnode7-012:0/16] 2024-12-05 18:30:06,109 (deepspeed_trainer:228) INFO: 13epoch:train:13601-13700batch: iter_time=1.138e-04, loss_ctc=78.059, loss_att=63.318, acc=0.661, loss=67.703, grad_norm=4.252, loss_scale=1.000, learning_rate=1.113e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-05 18:30:42,166 (deepspeed_trainer:228) INFO: 13epoch:train:13701-13800batch: iter_time=1.115e-04, loss_ctc=92.108, loss_att=68.256, acc=0.671, loss=75.408, grad_norm=4.599, loss_scale=1.000, learning_rate=1.113e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 18:31:18,215 (deepspeed_trainer:228) INFO: 13epoch:train:13801-13900batch: iter_time=1.136e-04, loss_ctc=80.851, loss_att=62.785, acc=0.668, loss=68.196, grad_norm=4.357, loss_scale=1.000, learning_rate=1.113e-04, step_time=0.360 [2024-12-05 18:31:54,380] [INFO] [logging.py:129:log_dist] [Rank 0] step=194000, skipped=0, lr=[np.float64(0.0001112253130012013)], mom=[[0.9, 0.98]] [2024-12-05 18:31:54,380] [INFO] [timer.py:264:stop] epoch=0/micro_step=14000/global_step=14000, RunningAvgSamplesPerSec=45.10758676247304, CurrSamplesPerSec=45.175815899626166, MemAllocated=1.77GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 18:31:54,382 (deepspeed_trainer:228) INFO: 13epoch:train:13901-14000batch: iter_time=1.094e-04, loss_ctc=88.029, loss_att=75.679, acc=0.661, loss=79.388, grad_norm=4.615, loss_scale=1.000, learning_rate=1.112e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 18:32:30,376 (deepspeed_trainer:228) INFO: 13epoch:train:14001-14100batch: iter_time=1.128e-04, loss_ctc=86.313, loss_att=65.936, acc=0.667, loss=72.041, grad_norm=4.787, loss_scale=1.000, learning_rate=1.112e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 18:33:06,165 (deepspeed_trainer:228) INFO: 13epoch:train:14101-14200batch: iter_time=1.194e-04, loss_ctc=85.783, loss_att=62.545, acc=0.678, loss=69.508, grad_norm=4.510, loss_scale=1.000, learning_rate=1.112e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-05 18:33:41,876 (deepspeed_trainer:228) INFO: 13epoch:train:14201-14300batch: iter_time=1.135e-04, loss_ctc=73.573, loss_att=60.941, acc=0.665, loss=64.749, grad_norm=3.897, loss_scale=1.000, learning_rate=1.112e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-05 18:34:17,790 (deepspeed_trainer:228) INFO: 13epoch:train:14301-14400batch: iter_time=1.141e-04, loss_ctc=93.775, loss_att=68.856, acc=0.665, loss=76.367, grad_norm=5.163, loss_scale=1.000, learning_rate=1.111e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 18:34:53,589 (deepspeed_trainer:228) INFO: 13epoch:train:14401-14500batch: iter_time=1.128e-04, loss_ctc=84.768, loss_att=66.443, acc=0.655, loss=71.914, grad_norm=5.059, loss_scale=1.000, learning_rate=1.111e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-05 18:35:29,773 (deepspeed_trainer:228) INFO: 13epoch:train:14501-14600batch: iter_time=1.159e-04, loss_ctc=75.193, loss_att=60.353, acc=0.668, loss=64.795, grad_norm=3.889, loss_scale=1.000, learning_rate=1.111e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 18:36:05,727 (deepspeed_trainer:228) INFO: 13epoch:train:14601-14700batch: iter_time=1.175e-04, loss_ctc=80.605, loss_att=66.362, acc=0.676, loss=70.633, grad_norm=3.746, loss_scale=1.000, learning_rate=1.110e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 18:36:41,866 (deepspeed_trainer:228) INFO: 13epoch:train:14701-14800batch: iter_time=1.163e-04, loss_ctc=84.414, loss_att=62.080, acc=0.680, loss=68.797, grad_norm=4.328, loss_scale=1.000, learning_rate=1.110e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 18:37:17,942 (deepspeed_trainer:228) INFO: 13epoch:train:14801-14900batch: iter_time=1.119e-04, loss_ctc=90.598, loss_att=67.284, acc=0.670, loss=74.258, grad_norm=4.438, loss_scale=1.000, learning_rate=1.110e-04, step_time=0.360 [2024-12-05 18:37:53,598] [INFO] [logging.py:129:log_dist] [Rank 0] step=195000, skipped=0, lr=[np.float64(0.00011093975478450083)], mom=[[0.9, 0.98]] [2024-12-05 18:37:53,599] [INFO] [timer.py:264:stop] epoch=0/micro_step=15000/global_step=15000, RunningAvgSamplesPerSec=45.15268204069895, CurrSamplesPerSec=49.08991153397351, MemAllocated=1.77GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 18:37:53,600 (deepspeed_trainer:228) INFO: 13epoch:train:14901-15000batch: iter_time=1.096e-04, loss_ctc=71.264, loss_att=52.306, acc=0.679, loss=57.988, grad_norm=4.426, loss_scale=1.000, learning_rate=1.110e-04, step_time=0.356 [2024-12-05 18:38:08,016] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:08,785] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:08,650] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:08,941] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:09,073] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:09,139] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:09,048] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:09,221] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:09,288] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:09,473] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:09,653] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:09,498] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:09,702] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:09,558] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:09,636] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:09,676] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:23,758] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:24,157] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:24,768] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:24,898] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:25,635] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:25,653] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:25,528] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:26,060] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:26,360] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:26,215] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:26,470] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:26,514] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:26,348] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:26,678] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:26,731] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:26,756] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:39,211] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:40,188] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:40,544] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:40,625] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:40,905] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:40,996] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:41,412] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:41,454] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:41,890] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:42,469] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:42,532] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:42,427] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:42,693] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:42,863] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:43,038] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:43,058] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:53,793] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:55,577] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:56,154] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:56,052] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:56,383] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:56,989] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:57,154] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:57,032] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:57,420] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:57,886] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:58,253] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:58,130] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:58,532] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:58,409] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:58,773] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:38:59,052] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:39:11,323] [INFO] [logging.py:129:log_dist] [Rank 0] [Torch] Checkpoint 13 is about to be saved! [2024-12-05 18:39:11,359] [INFO] [logging.py:129:log_dist] [Rank 0] Saving model checkpoint: exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/mp_rank_00_model_states.pt [2024-12-05 18:39:11,359] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/mp_rank_00_model_states.pt... [2024-12-05 18:39:13,157] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/mp_rank_00_model_states.pt. [2024-12-05 18:39:13,484] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_14_mp_rank_00_optim_states.pt... [2024-12-05 18:39:13,317] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt... [2024-12-05 18:39:13,321] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt... [2024-12-05 18:39:13,321] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt... [2024-12-05 18:39:13,321] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt... [2024-12-05 18:39:13,321] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt... [2024-12-05 18:39:13,322] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt... [2024-12-05 18:39:13,491] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_13_mp_rank_00_optim_states.pt... [2024-12-05 18:39:13,492] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_15_mp_rank_00_optim_states.pt... [2024-12-05 18:39:13,492] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_12_mp_rank_00_optim_states.pt... [2024-12-05 18:39:13,493] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_9_mp_rank_00_optim_states.pt... [2024-12-05 18:39:13,324] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt... [2024-12-05 18:39:13,493] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_11_mp_rank_00_optim_states.pt... [2024-12-05 18:39:13,493] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_8_mp_rank_00_optim_states.pt... [2024-12-05 18:39:13,325] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt... [2024-12-05 18:39:13,494] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_10_mp_rank_00_optim_states.pt... [2024-12-05 18:39:13,978] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt. [2024-12-05 18:39:13,978] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt [2024-12-05 18:39:13,978] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 13 is ready now! [2024-12-05 18:39:14,078] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt. [2024-12-05 18:39:14,078] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt [2024-12-05 18:39:14,078] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 13 is ready now! [2024-12-05 18:39:14,079] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt. [2024-12-05 18:39:14,087] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt [2024-12-05 18:39:14,087] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 13 is ready now! [2024-12-05 18:39:14,366] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_12_mp_rank_00_optim_states.pt. [2024-12-05 18:39:14,367] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_12_mp_rank_00_optim_states.pt [2024-12-05 18:39:14,367] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 13 is ready now! [2024-12-05 18:39:14,211] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt. [2024-12-05 18:39:14,211] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt [2024-12-05 18:39:14,211] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 13 is ready now! [2024-12-05 18:39:15,011] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt. [2024-12-05 18:39:15,011] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt [2024-12-05 18:39:15,011] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 13 is ready now! [2024-12-05 18:39:15,218] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_9_mp_rank_00_optim_states.pt. [2024-12-05 18:39:15,218] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_9_mp_rank_00_optim_states.pt [2024-12-05 18:39:15,218] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 13 is ready now! [2024-12-05 18:39:15,244] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_14_mp_rank_00_optim_states.pt. [2024-12-05 18:39:15,245] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_14_mp_rank_00_optim_states.pt [2024-12-05 18:39:15,245] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 13 is ready now! [2024-12-05 18:39:15,610] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_10_mp_rank_00_optim_states.pt. [2024-12-05 18:39:15,611] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_10_mp_rank_00_optim_states.pt [2024-12-05 18:39:15,611] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 13 is ready now! [2024-12-05 18:39:15,646] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_11_mp_rank_00_optim_states.pt. [2024-12-05 18:39:15,646] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_11_mp_rank_00_optim_states.pt [2024-12-05 18:39:15,646] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 13 is ready now! [2024-12-05 18:39:15,854] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_8_mp_rank_00_optim_states.pt. [2024-12-05 18:39:15,854] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_8_mp_rank_00_optim_states.pt [2024-12-05 18:39:15,854] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 13 is ready now! [2024-12-05 18:39:15,856] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_15_mp_rank_00_optim_states.pt. [2024-12-05 18:39:15,856] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_15_mp_rank_00_optim_states.pt [2024-12-05 18:39:15,856] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 13 is ready now! [2024-12-05 18:39:15,860] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_13_mp_rank_00_optim_states.pt. [2024-12-05 18:39:15,860] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_13_mp_rank_00_optim_states.pt [2024-12-05 18:39:15,860] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 13 is ready now! [2024-12-05 18:39:15,759] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt. [2024-12-05 18:39:15,759] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt [2024-12-05 18:39:15,759] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 13 is ready now! [2024-12-05 18:39:16,091] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt. [2024-12-05 18:39:16,091] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt [2024-12-05 18:39:16,091] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 13 is ready now! [2024-12-05 18:39:16,123] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt. [2024-12-05 18:39:16,123] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_13/13/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt [2024-12-05 18:39:16,123] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 13 is ready now! [cnode7-012:0/16] 2024-12-05 18:39:16,214 (deepspeed_trainer:158) INFO: 13epoch results: [train] iter_time=0.129, loss_ctc=83.779, loss_att=65.205, acc=0.666, loss=70.780, grad_norm=4.240, loss_scale=1.000, learning_rate=1.132e-04, step_time=0.365, time=2 hours, 3 minutes and 45.32 seconds, total_count=195013, gpu_max_cached_mem_GB=36.545, [valid] loss_ctc=5.750, cer_ctc=0.145, loss_att=9.000, acc=0.793, cer=0.427, wer=1.000, loss=8.062, time=1 minute and 9.27 seconds, total_count=13, gpu_max_cached_mem_GB=36.545 [cnode7-012:0/16] 2024-12-05 18:39:19,607 (multiple_iter_factory:32) INFO: Building 0th iter-factory... [cnode7-012:0/16] 2024-12-05 18:39:46,558 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-05 18:40:02,426 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.1", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.1", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.1", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.1", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-05 18:40:02,426 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.1, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-05 18:40:02,428 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-05 18:40:22,664] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:40:23,012] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:40:23,218] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:40:23,140] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:40:23,569] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:40:23,455] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:40:23,862] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:40:24,011] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:40:24,385] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:40:24,263] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:40:24,475] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:40:24,518] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:40:24,792] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:40:24,973] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:40:25,168] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:40:26,307] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:41:10,609] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:41:10,931] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:41:11,181] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:41:11,088] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:41:11,129] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:41:11,627] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:41:11,680] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:41:11,745] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:41:11,835] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:41:12,117] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:41:12,124] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:41:12,317] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:41:12,445] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:41:13,010] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:41:13,304] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:41:16,689] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:41:57,008] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:41:57,697] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:41:58,219] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:41:58,313] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:41:58,559] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:41:58,861] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:41:59,162] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:41:59,122] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:41:59,390] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:41:59,805] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:41:59,811] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:42:00,284] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:42:00,183] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:42:00,804] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:42:00,840] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:42:06,382] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:42:43,692] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:42:44,712] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:42:45,107] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:42:45,840] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:42:46,262] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:42:46,795] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:42:47,009] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:42:47,255] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:42:47,276] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:42:47,444] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:42:47,998] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:42:48,127] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:42:48,434] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:42:48,592] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:42:49,200] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:42:55,839] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) /mnt/home/williamchen/espnet/espnet2/s2t/espnet_model.py:279: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. with autocast(False): [cnode7-012:0/16] 2024-12-05 18:43:57,533 (deepspeed_trainer:228) INFO: 14epoch:train:1-100batch: iter_time=2.412, loss_ctc=88.724, loss_att=65.882, acc=0.664, loss=72.744, grad_norm=4.597, loss_scale=1.000, learning_rate=1.109e-04, step_time=0.367 [cnode7-012:0/16] 2024-12-05 18:44:33,743 (deepspeed_trainer:228) INFO: 14epoch:train:101-200batch: iter_time=1.117e-04, loss_ctc=83.006, loss_att=63.569, acc=0.667, loss=69.397, grad_norm=4.557, loss_scale=1.000, learning_rate=1.109e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 18:45:09,692 (deepspeed_trainer:228) INFO: 14epoch:train:201-300batch: iter_time=1.098e-04, loss_ctc=77.036, loss_att=56.672, acc=0.680, loss=62.779, grad_norm=3.923, loss_scale=1.000, learning_rate=1.109e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 18:45:45,948 (deepspeed_trainer:228) INFO: 14epoch:train:301-400batch: iter_time=1.141e-04, loss_ctc=90.541, loss_att=75.661, acc=0.639, loss=80.106, grad_norm=4.899, loss_scale=1.000, learning_rate=1.108e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 18:46:21,968 (deepspeed_trainer:228) INFO: 14epoch:train:401-500batch: iter_time=1.145e-04, loss_ctc=80.411, loss_att=63.830, acc=0.674, loss=68.808, grad_norm=4.019, loss_scale=1.000, learning_rate=1.108e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 18:46:57,781 (deepspeed_trainer:228) INFO: 14epoch:train:501-600batch: iter_time=1.103e-04, loss_ctc=71.109, loss_att=52.536, acc=0.689, loss=58.087, grad_norm=3.947, loss_scale=1.000, learning_rate=1.108e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 18:47:33,776 (deepspeed_trainer:228) INFO: 14epoch:train:601-700batch: iter_time=1.131e-04, loss_ctc=84.231, loss_att=66.536, acc=0.662, loss=71.853, grad_norm=5.095, loss_scale=1.000, learning_rate=1.108e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 18:48:09,590 (deepspeed_trainer:228) INFO: 14epoch:train:701-800batch: iter_time=1.124e-04, loss_ctc=89.811, loss_att=67.493, acc=0.665, loss=74.200, grad_norm=4.667, loss_scale=1.000, learning_rate=1.107e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 18:48:45,212 (deepspeed_trainer:228) INFO: 14epoch:train:801-900batch: iter_time=1.109e-04, loss_ctc=82.978, loss_att=58.424, acc=0.681, loss=65.797, grad_norm=4.637, loss_scale=1.000, learning_rate=1.107e-04, step_time=0.356 [2024-12-05 18:49:20,792] [INFO] [logging.py:129:log_dist] [Rank 0] step=196000, skipped=0, lr=[np.float64(0.0001106563847481619)], mom=[[0.9, 0.98]] [2024-12-05 18:49:20,792] [INFO] [timer.py:264:stop] epoch=0/micro_step=16000/global_step=16000, RunningAvgSamplesPerSec=45.188455770695676, CurrSamplesPerSec=47.25230303014772, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 18:49:20,794 (deepspeed_trainer:228) INFO: 14epoch:train:901-1000batch: iter_time=1.085e-04, loss_ctc=74.362, loss_att=55.523, acc=0.672, loss=61.157, grad_norm=4.132, loss_scale=1.000, learning_rate=1.107e-04, step_time=0.356 [cnode7-012:0/16] 2024-12-05 18:49:56,419 (deepspeed_trainer:228) INFO: 14epoch:train:1001-1100batch: iter_time=1.097e-04, loss_ctc=76.905, loss_att=62.012, acc=0.665, loss=66.457, grad_norm=3.922, loss_scale=1.000, learning_rate=1.106e-04, step_time=0.356 [cnode7-012:0/16] 2024-12-05 18:50:32,080 (deepspeed_trainer:228) INFO: 14epoch:train:1101-1200batch: iter_time=1.106e-04, loss_ctc=82.351, loss_att=67.244, acc=0.668, loss=71.772, grad_norm=4.310, loss_scale=1.000, learning_rate=1.106e-04, step_time=0.356 [cnode7-012:0/16] 2024-12-05 18:51:08,031 (deepspeed_trainer:228) INFO: 14epoch:train:1201-1300batch: iter_time=1.133e-04, loss_ctc=84.976, loss_att=63.351, acc=0.664, loss=69.820, grad_norm=4.599, loss_scale=1.000, learning_rate=1.106e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 18:51:43,945 (deepspeed_trainer:228) INFO: 14epoch:train:1301-1400batch: iter_time=1.119e-04, loss_ctc=81.901, loss_att=60.737, acc=0.686, loss=67.054, grad_norm=4.646, loss_scale=1.000, learning_rate=1.106e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 18:52:20,358 (deepspeed_trainer:228) INFO: 14epoch:train:1401-1500batch: iter_time=1.092e-04, loss_ctc=80.175, loss_att=55.373, acc=0.681, loss=62.819, grad_norm=4.356, loss_scale=1.000, learning_rate=1.105e-04, step_time=0.364 [cnode7-012:0/16] 2024-12-05 18:52:56,425 (deepspeed_trainer:228) INFO: 14epoch:train:1501-1600batch: iter_time=1.119e-04, loss_ctc=79.796, loss_att=62.604, acc=0.683, loss=67.751, grad_norm=4.152, loss_scale=1.000, learning_rate=1.105e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 18:53:32,363 (deepspeed_trainer:228) INFO: 14epoch:train:1601-1700batch: iter_time=1.101e-04, loss_ctc=77.697, loss_att=68.336, acc=0.665, loss=71.156, grad_norm=4.150, loss_scale=1.000, learning_rate=1.105e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 18:54:07,938 (deepspeed_trainer:228) INFO: 14epoch:train:1701-1800batch: iter_time=1.123e-04, loss_ctc=75.609, loss_att=54.300, acc=0.678, loss=60.706, grad_norm=3.641, loss_scale=1.000, learning_rate=1.104e-04, step_time=0.355 [cnode7-012:0/16] 2024-12-05 18:54:38,171 (multiple_iter_factory:32) INFO: Building 1th iter-factory... [cnode7-012:0/16] 2024-12-05 18:55:03,978 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-05 18:55:18,972 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.7", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.7", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.7", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.7", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-05 18:55:18,972 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.7, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-05 18:55:18,974 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-05 18:55:49,249] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:55:49,307] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:55:49,240] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:55:49,396] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:55:49,470] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:55:49,479] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:55:49,887] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:55:49,909] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:55:49,931] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:55:50,486] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:55:50,394] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:55:50,461] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:55:50,802] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:55:50,842] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:55:51,819] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:55:52,674] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:56:40,511] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:56:40,992] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:56:41,025] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:56:40,916] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:56:41,994] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:56:41,895] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:56:41,920] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:56:42,239] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:56:42,307] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:56:42,314] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:56:42,601] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:56:42,756] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:56:42,917] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:56:43,034] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:56:43,620] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:56:44,559] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:57:29,277] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:57:29,484] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:57:30,074] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:57:30,102] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:57:30,106] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:57:30,374] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:57:30,610] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:57:30,826] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:57:31,048] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:57:31,313] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:57:31,623] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:57:32,252] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:57:32,584] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:57:34,619] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:57:35,473] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:57:35,514] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:58:16,991] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:58:17,055] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:58:17,541] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:58:18,439] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:58:18,587] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:58:18,631] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:58:18,887] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:58:19,078] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:58:19,354] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:58:19,468] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:58:19,681] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:58:19,780] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:58:23,434] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:58:24,389] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:58:25,934] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 18:58:28,762] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-05 18:59:03,123 (deepspeed_trainer:228) INFO: 14epoch:train:1801-1900batch: iter_time=2.481, loss_ctc=77.266, loss_att=57.801, acc=0.663, loss=63.606, grad_norm=4.309, loss_scale=1.000, learning_rate=1.104e-04, step_time=0.470 [2024-12-05 18:59:39,524] [INFO] [logging.py:129:log_dist] [Rank 0] step=197000, skipped=0, lr=[np.float64(0.00011037517508814978)], mom=[[0.9, 0.98]] [2024-12-05 18:59:39,524] [INFO] [timer.py:264:stop] epoch=0/micro_step=17000/global_step=17000, RunningAvgSamplesPerSec=45.138614167130214, CurrSamplesPerSec=47.892784768707195, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 18:59:39,526 (deepspeed_trainer:228) INFO: 14epoch:train:1901-2000batch: iter_time=1.112e-04, loss_ctc=88.619, loss_att=66.282, acc=0.668, loss=73.026, grad_norm=4.597, loss_scale=1.000, learning_rate=1.104e-04, step_time=0.364 [cnode7-012:0/16] 2024-12-05 19:00:15,627 (deepspeed_trainer:228) INFO: 14epoch:train:2001-2100batch: iter_time=1.130e-04, loss_ctc=78.235, loss_att=60.070, acc=0.669, loss=65.523, grad_norm=4.325, loss_scale=1.000, learning_rate=1.104e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 19:00:51,478 (deepspeed_trainer:228) INFO: 14epoch:train:2101-2200batch: iter_time=1.107e-04, loss_ctc=77.753, loss_att=57.036, acc=0.687, loss=63.218, grad_norm=3.920, loss_scale=1.000, learning_rate=1.103e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 19:01:27,847 (deepspeed_trainer:228) INFO: 14epoch:train:2201-2300batch: iter_time=1.156e-04, loss_ctc=85.561, loss_att=72.444, acc=0.638, loss=76.362, grad_norm=4.356, loss_scale=1.000, learning_rate=1.103e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-05 19:02:03,964 (deepspeed_trainer:228) INFO: 14epoch:train:2301-2400batch: iter_time=1.116e-04, loss_ctc=79.839, loss_att=60.953, acc=0.688, loss=66.632, grad_norm=3.808, loss_scale=1.000, learning_rate=1.103e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 19:02:40,254 (deepspeed_trainer:228) INFO: 14epoch:train:2401-2500batch: iter_time=1.092e-04, loss_ctc=75.037, loss_att=57.352, acc=0.682, loss=62.638, grad_norm=4.324, loss_scale=1.000, learning_rate=1.102e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-05 19:03:16,649 (deepspeed_trainer:228) INFO: 14epoch:train:2501-2600batch: iter_time=1.110e-04, loss_ctc=93.472, loss_att=73.824, acc=0.652, loss=79.720, grad_norm=5.148, loss_scale=1.000, learning_rate=1.102e-04, step_time=0.364 [cnode7-012:0/16] 2024-12-05 19:03:52,719 (deepspeed_trainer:228) INFO: 14epoch:train:2601-2700batch: iter_time=1.111e-04, loss_ctc=76.063, loss_att=57.203, acc=0.686, loss=62.849, grad_norm=3.861, loss_scale=1.000, learning_rate=1.102e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 19:04:28,710 (deepspeed_trainer:228) INFO: 14epoch:train:2701-2800batch: iter_time=1.120e-04, loss_ctc=83.529, loss_att=56.814, acc=0.676, loss=64.806, grad_norm=4.560, loss_scale=1.000, learning_rate=1.102e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 19:05:05,251 (deepspeed_trainer:228) INFO: 14epoch:train:2801-2900batch: iter_time=1.096e-04, loss_ctc=70.104, loss_att=52.905, acc=0.682, loss=58.067, grad_norm=3.882, loss_scale=1.000, learning_rate=1.101e-04, step_time=0.365 [2024-12-05 19:05:41,429] [INFO] [logging.py:129:log_dist] [Rank 0] step=198000, skipped=0, lr=[np.float64(0.00011009609849253636)], mom=[[0.9, 0.98]] [2024-12-05 19:05:41,429] [INFO] [timer.py:264:stop] epoch=0/micro_step=18000/global_step=18000, RunningAvgSamplesPerSec=45.15445148961856, CurrSamplesPerSec=47.24658110822637, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 19:05:41,431 (deepspeed_trainer:228) INFO: 14epoch:train:2901-3000batch: iter_time=1.102e-04, loss_ctc=78.071, loss_att=64.648, acc=0.665, loss=68.688, grad_norm=4.167, loss_scale=1.000, learning_rate=1.101e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 19:06:17,473 (deepspeed_trainer:228) INFO: 14epoch:train:3001-3100batch: iter_time=1.155e-04, loss_ctc=80.030, loss_att=61.843, acc=0.675, loss=67.297, grad_norm=3.959, loss_scale=1.000, learning_rate=1.101e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 19:06:53,347 (deepspeed_trainer:228) INFO: 14epoch:train:3101-3200batch: iter_time=1.121e-04, loss_ctc=85.148, loss_att=63.922, acc=0.671, loss=70.309, grad_norm=5.145, loss_scale=1.000, learning_rate=1.101e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 19:07:29,239 (deepspeed_trainer:228) INFO: 14epoch:train:3201-3300batch: iter_time=1.104e-04, loss_ctc=82.113, loss_att=60.094, acc=0.688, loss=66.699, grad_norm=4.303, loss_scale=1.000, learning_rate=1.100e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 19:08:05,133 (deepspeed_trainer:228) INFO: 14epoch:train:3301-3400batch: iter_time=1.212e-04, loss_ctc=76.903, loss_att=57.046, acc=0.685, loss=63.010, grad_norm=4.247, loss_scale=1.000, learning_rate=1.100e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 19:08:41,074 (deepspeed_trainer:228) INFO: 14epoch:train:3401-3500batch: iter_time=1.096e-04, loss_ctc=78.331, loss_att=58.218, acc=0.680, loss=64.248, grad_norm=3.916, loss_scale=1.000, learning_rate=1.100e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 19:09:17,165 (deepspeed_trainer:228) INFO: 14epoch:train:3501-3600batch: iter_time=1.106e-04, loss_ctc=76.922, loss_att=65.748, acc=0.675, loss=69.126, grad_norm=4.106, loss_scale=1.000, learning_rate=1.099e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 19:09:53,401 (deepspeed_trainer:228) INFO: 14epoch:train:3601-3700batch: iter_time=1.103e-04, loss_ctc=76.829, loss_att=54.588, acc=0.683, loss=61.273, grad_norm=3.838, loss_scale=1.000, learning_rate=1.099e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 19:10:13,902 (multiple_iter_factory:32) INFO: Building 2th iter-factory... [cnode7-012:0/16] 2024-12-05 19:10:39,839 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-05 19:10:55,797 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.6", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.6", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.6", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.6", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-05 19:10:55,797 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.6, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-05 19:10:55,799 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-05 19:11:22,057] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:11:22,373] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:11:22,584] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:11:22,748] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:11:24,458] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:11:24,736] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:11:24,913] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:11:25,067] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:11:25,217] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:11:25,227] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:11:25,369] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:11:25,792] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:11:25,708] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:11:25,915] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:11:26,094] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:11:26,286] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:12:09,749] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:12:10,235] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:12:10,801] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:12:11,861] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:12:12,614] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:12:12,671] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:12:13,151] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:12:13,673] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:12:13,654] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:12:13,838] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:12:14,058] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:12:14,270] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:12:14,439] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:12:14,732] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:12:14,956] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:12:15,117] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:12:57,760] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:12:57,972] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:12:58,007] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:13:00,664] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:13:00,958] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:13:01,195] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:13:01,552] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:13:02,177] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:13:02,532] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:13:02,548] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:13:02,741] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:13:02,790] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:13:03,374] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:13:03,552] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:13:03,989] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:13:06,578] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:13:44,535] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:13:44,739] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:13:45,497] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:13:47,328] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:13:48,040] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:13:49,344] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:13:49,588] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:13:49,630] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:13:49,993] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:13:50,511] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:13:50,860] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:13:50,783] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:13:51,187] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:13:51,426] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:13:55,000] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:13:56,127] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-05 19:14:39,903 (deepspeed_trainer:228) INFO: 14epoch:train:3701-3800batch: iter_time=2.418, loss_ctc=77.966, loss_att=64.338, acc=0.657, loss=68.446, grad_norm=4.444, loss_scale=1.000, learning_rate=1.099e-04, step_time=0.447 [cnode7-012:0/16] 2024-12-05 19:15:15,946 (deepspeed_trainer:228) INFO: 14epoch:train:3801-3900batch: iter_time=1.089e-04, loss_ctc=84.879, loss_att=66.720, acc=0.674, loss=72.164, grad_norm=4.508, loss_scale=1.000, learning_rate=1.099e-04, step_time=0.360 [2024-12-05 19:15:52,726] [INFO] [logging.py:129:log_dist] [Rank 0] step=199000, skipped=0, lr=[np.float64(0.00010981912813035827)], mom=[[0.9, 0.98]] [2024-12-05 19:15:52,727] [INFO] [timer.py:264:stop] epoch=0/micro_step=19000/global_step=19000, RunningAvgSamplesPerSec=45.11738155517088, CurrSamplesPerSec=45.308951554370985, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 19:15:52,728 (deepspeed_trainer:228) INFO: 14epoch:train:3901-4000batch: iter_time=1.096e-04, loss_ctc=79.704, loss_att=64.824, acc=0.667, loss=69.289, grad_norm=3.986, loss_scale=1.000, learning_rate=1.098e-04, step_time=0.367 [cnode7-012:0/16] 2024-12-05 19:16:29,035 (deepspeed_trainer:228) INFO: 14epoch:train:4001-4100batch: iter_time=1.112e-04, loss_ctc=76.956, loss_att=64.405, acc=0.679, loss=68.140, grad_norm=3.934, loss_scale=1.000, learning_rate=1.098e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-05 19:17:05,195 (deepspeed_trainer:228) INFO: 14epoch:train:4101-4200batch: iter_time=1.122e-04, loss_ctc=85.417, loss_att=71.317, acc=0.653, loss=75.579, grad_norm=4.499, loss_scale=1.000, learning_rate=1.098e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 19:17:41,211 (deepspeed_trainer:228) INFO: 14epoch:train:4201-4300batch: iter_time=1.103e-04, loss_ctc=78.046, loss_att=61.464, acc=0.685, loss=66.391, grad_norm=4.011, loss_scale=1.000, learning_rate=1.098e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 19:18:17,030 (deepspeed_trainer:228) INFO: 14epoch:train:4301-4400batch: iter_time=1.113e-04, loss_ctc=74.623, loss_att=56.978, acc=0.690, loss=62.264, grad_norm=4.112, loss_scale=1.000, learning_rate=1.097e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 19:18:53,669 (deepspeed_trainer:228) INFO: 14epoch:train:4401-4500batch: iter_time=1.106e-04, loss_ctc=92.644, loss_att=70.993, acc=0.661, loss=77.510, grad_norm=5.121, loss_scale=1.000, learning_rate=1.097e-04, step_time=0.366 [cnode7-012:0/16] 2024-12-05 19:19:29,699 (deepspeed_trainer:228) INFO: 14epoch:train:4501-4600batch: iter_time=1.137e-04, loss_ctc=73.097, loss_att=55.925, acc=0.692, loss=61.078, grad_norm=4.163, loss_scale=1.000, learning_rate=1.097e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 19:20:05,389 (deepspeed_trainer:228) INFO: 14epoch:train:4601-4700batch: iter_time=1.118e-04, loss_ctc=80.513, loss_att=58.554, acc=0.680, loss=65.159, grad_norm=4.780, loss_scale=1.000, learning_rate=1.096e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-05 19:20:41,166 (deepspeed_trainer:228) INFO: 14epoch:train:4701-4800batch: iter_time=1.121e-04, loss_ctc=75.543, loss_att=59.737, acc=0.678, loss=64.481, grad_norm=5.074, loss_scale=1.000, learning_rate=1.096e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-05 19:21:17,017 (deepspeed_trainer:228) INFO: 14epoch:train:4801-4900batch: iter_time=1.125e-04, loss_ctc=75.925, loss_att=62.476, acc=0.674, loss=66.501, grad_norm=4.284, loss_scale=1.000, learning_rate=1.096e-04, step_time=0.358 [2024-12-05 19:21:52,887] [INFO] [logging.py:129:log_dist] [Rank 0] step=200000, skipped=0, lr=[np.float64(0.00010954423764078145)], mom=[[0.9, 0.98]] [2024-12-05 19:21:52,887] [INFO] [timer.py:264:stop] epoch=0/micro_step=20000/global_step=20000, RunningAvgSamplesPerSec=45.145520613215574, CurrSamplesPerSec=42.159020813714775, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 19:21:52,889 (deepspeed_trainer:228) INFO: 14epoch:train:4901-5000batch: iter_time=1.106e-04, loss_ctc=83.416, loss_att=63.658, acc=0.680, loss=69.570, grad_norm=4.336, loss_scale=1.000, learning_rate=1.096e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 19:22:29,055 (deepspeed_trainer:228) INFO: 14epoch:train:5001-5100batch: iter_time=1.127e-04, loss_ctc=82.141, loss_att=64.233, acc=0.679, loss=69.593, grad_norm=5.073, loss_scale=1.000, learning_rate=1.095e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 19:23:04,631 (deepspeed_trainer:228) INFO: 14epoch:train:5101-5200batch: iter_time=1.172e-04, loss_ctc=80.174, loss_att=58.494, acc=0.693, loss=65.013, grad_norm=3.975, loss_scale=1.000, learning_rate=1.095e-04, step_time=0.356 [cnode7-012:0/16] 2024-12-05 19:23:40,260 (deepspeed_trainer:228) INFO: 14epoch:train:5201-5300batch: iter_time=1.212e-04, loss_ctc=82.356, loss_att=62.133, acc=0.686, loss=68.232, grad_norm=4.223, loss_scale=1.000, learning_rate=1.095e-04, step_time=0.356 [cnode7-012:0/16] 2024-12-05 19:24:15,906 (deepspeed_trainer:228) INFO: 14epoch:train:5301-5400batch: iter_time=1.176e-04, loss_ctc=73.945, loss_att=63.421, acc=0.681, loss=66.579, grad_norm=3.977, loss_scale=1.000, learning_rate=1.094e-04, step_time=0.356 [cnode7-012:0/16] 2024-12-05 19:24:51,345 (deepspeed_trainer:228) INFO: 14epoch:train:5401-5500batch: iter_time=1.199e-04, loss_ctc=75.017, loss_att=56.820, acc=0.691, loss=62.300, grad_norm=3.994, loss_scale=1.000, learning_rate=1.094e-04, step_time=0.354 [cnode7-012:0/16] 2024-12-05 19:25:26,969 (deepspeed_trainer:228) INFO: 14epoch:train:5501-5600batch: iter_time=1.188e-04, loss_ctc=75.340, loss_att=59.094, acc=0.668, loss=63.959, grad_norm=3.902, loss_scale=1.000, learning_rate=1.094e-04, step_time=0.356 [cnode7-012:0/16] 2024-12-05 19:25:38,328 (multiple_iter_factory:32) INFO: Building 3th iter-factory... [cnode7-012:0/16] 2024-12-05 19:26:04,331 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-05 19:26:20,937 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.3", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.3", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.3", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.3", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-05 19:26:20,937 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.3, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-05 19:26:20,939 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-05 19:26:46,503] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:26:47,460] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:26:47,658] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:26:48,036] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:26:48,427] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:26:48,674] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:26:48,653] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:26:49,864] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:26:50,296] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:26:50,430] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:26:50,592] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:26:50,635] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:26:50,940] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:26:50,989] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:26:51,249] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:26:53,055] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:27:34,276] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:27:35,008] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:27:35,396] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:27:35,847] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:27:36,275] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:27:37,425] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:27:37,584] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:27:39,013] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:27:39,106] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:27:39,169] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:27:39,401] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:27:40,035] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:27:40,228] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:27:40,529] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:27:40,576] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:27:41,094] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:28:20,794] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:28:21,705] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:28:22,889] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:28:23,668] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:28:24,490] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:28:26,386] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:28:26,698] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:28:27,461] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:28:27,724] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:28:28,283] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:28:28,308] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:28:28,579] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:28:29,095] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:28:29,194] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:28:29,467] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:28:29,577] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:29:07,643] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:29:08,730] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:29:09,193] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:29:11,457] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:29:12,912] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:29:14,751] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:29:15,082] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:29:15,840] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:29:16,427] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:29:16,742] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:29:17,321] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:29:17,336] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:29:17,470] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:29:17,609] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:29:17,866] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:29:20,795] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-05 19:30:14,367 (deepspeed_trainer:228) INFO: 14epoch:train:5601-5700batch: iter_time=2.426, loss_ctc=80.459, loss_att=61.829, acc=0.674, loss=67.449, grad_norm=4.343, loss_scale=1.000, learning_rate=1.094e-04, step_time=0.448 [cnode7-012:0/16] 2024-12-05 19:30:50,508 (deepspeed_trainer:228) INFO: 14epoch:train:5701-5800batch: iter_time=1.078e-04, loss_ctc=84.789, loss_att=62.915, acc=0.664, loss=69.476, grad_norm=4.700, loss_scale=1.000, learning_rate=1.093e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 19:31:26,970 (deepspeed_trainer:228) INFO: 14epoch:train:5801-5900batch: iter_time=1.132e-04, loss_ctc=77.047, loss_att=59.350, acc=0.681, loss=64.676, grad_norm=3.831, loss_scale=1.000, learning_rate=1.093e-04, step_time=0.364 [2024-12-05 19:32:03,422] [INFO] [logging.py:129:log_dist] [Rank 0] step=201000, skipped=0, lr=[np.float64(0.0001092714011225629)], mom=[[0.9, 0.98]] [2024-12-05 19:32:03,424] [INFO] [timer.py:264:stop] epoch=0/micro_step=21000/global_step=21000, RunningAvgSamplesPerSec=45.123155299879215, CurrSamplesPerSec=47.559583202685346, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 19:32:03,426 (deepspeed_trainer:228) INFO: 14epoch:train:5901-6000batch: iter_time=1.743e-04, loss_ctc=81.069, loss_att=65.434, acc=0.670, loss=70.122, grad_norm=4.307, loss_scale=1.000, learning_rate=1.093e-04, step_time=0.364 [cnode7-012:0/16] 2024-12-05 19:32:40,284 (deepspeed_trainer:228) INFO: 14epoch:train:6001-6100batch: iter_time=1.128e-04, loss_ctc=88.094, loss_att=69.043, acc=0.661, loss=74.780, grad_norm=4.322, loss_scale=1.000, learning_rate=1.093e-04, step_time=0.368 [cnode7-012:0/16] 2024-12-05 19:33:17,252 (deepspeed_trainer:228) INFO: 14epoch:train:6101-6200batch: iter_time=1.132e-04, loss_ctc=71.933, loss_att=56.169, acc=0.690, loss=60.888, grad_norm=3.793, loss_scale=1.000, learning_rate=1.092e-04, step_time=0.369 [cnode7-012:0/16] 2024-12-05 19:33:53,399 (deepspeed_trainer:228) INFO: 14epoch:train:6201-6300batch: iter_time=1.153e-04, loss_ctc=77.528, loss_att=61.542, acc=0.681, loss=66.317, grad_norm=4.331, loss_scale=1.000, learning_rate=1.092e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 19:34:29,326 (deepspeed_trainer:228) INFO: 14epoch:train:6301-6400batch: iter_time=1.104e-04, loss_ctc=86.190, loss_att=62.553, acc=0.671, loss=69.621, grad_norm=4.484, loss_scale=1.000, learning_rate=1.092e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 19:35:04,844 (deepspeed_trainer:228) INFO: 14epoch:train:6401-6500batch: iter_time=1.115e-04, loss_ctc=75.274, loss_att=54.838, acc=0.687, loss=60.950, grad_norm=4.155, loss_scale=1.000, learning_rate=1.091e-04, step_time=0.355 [cnode7-012:0/16] 2024-12-05 19:35:41,006 (deepspeed_trainer:228) INFO: 14epoch:train:6501-6600batch: iter_time=1.154e-04, loss_ctc=79.558, loss_att=59.017, acc=0.676, loss=65.187, grad_norm=4.287, loss_scale=1.000, learning_rate=1.091e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 19:36:17,011 (deepspeed_trainer:228) INFO: 14epoch:train:6601-6700batch: iter_time=1.120e-04, loss_ctc=74.166, loss_att=59.388, acc=0.675, loss=63.830, grad_norm=4.077, loss_scale=1.000, learning_rate=1.091e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 19:36:53,411 (deepspeed_trainer:228) INFO: 14epoch:train:6701-6800batch: iter_time=1.128e-04, loss_ctc=77.028, loss_att=62.684, acc=0.668, loss=67.009, grad_norm=4.182, loss_scale=1.000, learning_rate=1.091e-04, step_time=0.364 [cnode7-012:0/16] 2024-12-05 19:37:29,794 (deepspeed_trainer:228) INFO: 14epoch:train:6801-6900batch: iter_time=1.128e-04, loss_ctc=83.630, loss_att=62.994, acc=0.679, loss=69.163, grad_norm=4.751, loss_scale=1.000, learning_rate=1.090e-04, step_time=0.364 [2024-12-05 19:38:05,727] [INFO] [logging.py:129:log_dist] [Rank 0] step=202000, skipped=0, lr=[np.float64(0.00010900059312379953)], mom=[[0.9, 0.98]] [2024-12-05 19:38:05,728] [INFO] [timer.py:264:stop] epoch=0/micro_step=22000/global_step=22000, RunningAvgSamplesPerSec=45.13632722032414, CurrSamplesPerSec=42.996530391589786, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 19:38:05,729 (deepspeed_trainer:228) INFO: 14epoch:train:6901-7000batch: iter_time=1.113e-04, loss_ctc=81.351, loss_att=59.090, acc=0.690, loss=65.776, grad_norm=4.116, loss_scale=1.000, learning_rate=1.090e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 19:38:41,566 (deepspeed_trainer:228) INFO: 14epoch:train:7001-7100batch: iter_time=1.137e-04, loss_ctc=80.698, loss_att=58.949, acc=0.683, loss=65.479, grad_norm=4.131, loss_scale=1.000, learning_rate=1.090e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 19:39:17,356 (deepspeed_trainer:228) INFO: 14epoch:train:7101-7200batch: iter_time=1.135e-04, loss_ctc=77.801, loss_att=57.913, acc=0.691, loss=63.886, grad_norm=4.155, loss_scale=1.000, learning_rate=1.090e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 19:39:53,461 (deepspeed_trainer:228) INFO: 14epoch:train:7201-7300batch: iter_time=1.100e-04, loss_ctc=75.248, loss_att=66.511, acc=0.673, loss=69.159, grad_norm=4.118, loss_scale=1.000, learning_rate=1.089e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 19:40:29,331 (deepspeed_trainer:228) INFO: 14epoch:train:7301-7400batch: iter_time=1.143e-04, loss_ctc=77.175, loss_att=54.502, acc=0.684, loss=61.343, grad_norm=3.994, loss_scale=1.000, learning_rate=1.089e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 19:41:04,765 (deepspeed_trainer:228) INFO: 14epoch:train:7401-7500batch: iter_time=1.115e-04, loss_ctc=71.655, loss_att=55.201, acc=0.671, loss=60.141, grad_norm=4.119, loss_scale=1.000, learning_rate=1.089e-04, step_time=0.354 [cnode7-012:0/16] 2024-12-05 19:41:07,669 (multiple_iter_factory:32) INFO: Building 4th iter-factory... [cnode7-012:0/16] 2024-12-05 19:41:34,174 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-05 19:41:50,764 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.2", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.2", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.2", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.2", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-05 19:41:50,764 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.2, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-05 19:41:50,766 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-05 19:42:18,692] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:42:18,755] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:42:18,787] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:42:18,804] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:42:18,848] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:42:18,888] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:42:19,075] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:42:19,267] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:42:19,714] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:42:20,072] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:42:20,188] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:42:20,590] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:42:20,840] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:42:20,999] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:42:20,957] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:42:21,680] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:43:06,344] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:43:06,454] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:43:06,650] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:43:06,748] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:43:06,969] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:43:07,509] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:43:08,052] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:43:08,246] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:43:08,389] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:43:09,044] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:43:09,073] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:43:09,041] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:43:09,125] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:43:09,698] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:43:10,142] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:43:11,206] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:43:53,639] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:43:54,391] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:43:55,653] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:43:56,020] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:43:56,451] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:43:56,509] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:43:57,213] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:43:57,334] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:43:57,929] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:43:57,994] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:43:58,347] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:43:58,493] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:43:58,474] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:43:58,603] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:43:58,549] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:44:01,094] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:44:41,163] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:44:43,549] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:44:43,976] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:44:44,028] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:44:44,223] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:44:44,363] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:44:45,263] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:44:46,287] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:44:46,675] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:44:46,913] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:44:48,048] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:44:48,197] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:44:48,308] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:44:48,479] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:44:48,808] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:44:51,497] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-05 19:45:54,618 (deepspeed_trainer:228) INFO: 14epoch:train:7501-7600batch: iter_time=2.502, loss_ctc=86.323, loss_att=68.728, acc=0.670, loss=74.038, grad_norm=4.302, loss_scale=1.000, learning_rate=1.089e-04, step_time=0.396 [cnode7-012:0/16] 2024-12-05 19:46:31,501 (deepspeed_trainer:228) INFO: 14epoch:train:7601-7700batch: iter_time=1.092e-04, loss_ctc=80.566, loss_att=65.762, acc=0.670, loss=70.216, grad_norm=4.417, loss_scale=1.000, learning_rate=1.088e-04, step_time=0.369 [cnode7-012:0/16] 2024-12-05 19:47:08,119 (deepspeed_trainer:228) INFO: 14epoch:train:7701-7800batch: iter_time=1.099e-04, loss_ctc=73.487, loss_att=56.698, acc=0.689, loss=61.781, grad_norm=3.899, loss_scale=1.000, learning_rate=1.088e-04, step_time=0.366 [cnode7-012:0/16] 2024-12-05 19:47:44,962 (deepspeed_trainer:228) INFO: 14epoch:train:7801-7900batch: iter_time=1.070e-04, loss_ctc=88.277, loss_att=74.618, acc=0.654, loss=78.725, grad_norm=4.459, loss_scale=1.000, learning_rate=1.088e-04, step_time=0.368 [2024-12-05 19:48:21,228] [INFO] [logging.py:129:log_dist] [Rank 0] step=203000, skipped=0, lr=[np.float64(0.00010873178863195503)], mom=[[0.9, 0.98]] [2024-12-05 19:48:21,228] [INFO] [timer.py:264:stop] epoch=0/micro_step=23000/global_step=23000, RunningAvgSamplesPerSec=45.13114351921551, CurrSamplesPerSec=45.04515348405908, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 19:48:21,230 (deepspeed_trainer:228) INFO: 14epoch:train:7901-8000batch: iter_time=1.087e-04, loss_ctc=78.931, loss_att=62.767, acc=0.686, loss=67.620, grad_norm=4.162, loss_scale=1.000, learning_rate=1.087e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 19:48:57,357 (deepspeed_trainer:228) INFO: 14epoch:train:8001-8100batch: iter_time=1.102e-04, loss_ctc=70.144, loss_att=51.873, acc=0.697, loss=57.367, grad_norm=3.740, loss_scale=1.000, learning_rate=1.087e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 19:49:33,616 (deepspeed_trainer:228) INFO: 14epoch:train:8101-8200batch: iter_time=1.131e-04, loss_ctc=81.228, loss_att=65.027, acc=0.675, loss=69.875, grad_norm=4.402, loss_scale=1.000, learning_rate=1.087e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 19:50:09,835 (deepspeed_trainer:228) INFO: 14epoch:train:8201-8300batch: iter_time=1.098e-04, loss_ctc=86.289, loss_att=65.313, acc=0.680, loss=71.580, grad_norm=4.441, loss_scale=1.000, learning_rate=1.087e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 19:50:45,916 (deepspeed_trainer:228) INFO: 14epoch:train:8301-8400batch: iter_time=1.099e-04, loss_ctc=81.301, loss_att=58.305, acc=0.689, loss=65.174, grad_norm=4.644, loss_scale=1.000, learning_rate=1.086e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 19:51:21,937 (deepspeed_trainer:228) INFO: 14epoch:train:8401-8500batch: iter_time=1.069e-04, loss_ctc=72.574, loss_att=54.913, acc=0.683, loss=60.239, grad_norm=4.100, loss_scale=1.000, learning_rate=1.086e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 19:51:57,887 (deepspeed_trainer:228) INFO: 14epoch:train:8501-8600batch: iter_time=1.087e-04, loss_ctc=75.504, loss_att=61.987, acc=0.675, loss=66.074, grad_norm=4.001, loss_scale=1.000, learning_rate=1.086e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 19:52:33,920 (deepspeed_trainer:228) INFO: 14epoch:train:8601-8700batch: iter_time=1.083e-04, loss_ctc=80.088, loss_att=65.284, acc=0.682, loss=69.733, grad_norm=3.999, loss_scale=1.000, learning_rate=1.086e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 19:53:09,835 (deepspeed_trainer:228) INFO: 14epoch:train:8701-8800batch: iter_time=1.090e-04, loss_ctc=83.007, loss_att=62.508, acc=0.676, loss=68.663, grad_norm=4.634, loss_scale=1.000, learning_rate=1.085e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 19:53:45,844 (deepspeed_trainer:228) INFO: 14epoch:train:8801-8900batch: iter_time=1.095e-04, loss_ctc=79.384, loss_att=61.346, acc=0.693, loss=66.743, grad_norm=4.116, loss_scale=1.000, learning_rate=1.085e-04, step_time=0.360 [2024-12-05 19:54:21,654] [INFO] [logging.py:129:log_dist] [Rank 0] step=204000, skipped=0, lr=[np.float64(0.00010846496306415617)], mom=[[0.9, 0.98]] [2024-12-05 19:54:21,655] [INFO] [timer.py:264:stop] epoch=0/micro_step=24000/global_step=24000, RunningAvgSamplesPerSec=45.15211228568421, CurrSamplesPerSec=47.59682281831544, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 19:54:21,657 (deepspeed_trainer:228) INFO: 14epoch:train:8901-9000batch: iter_time=1.100e-04, loss_ctc=78.168, loss_att=54.371, acc=0.693, loss=61.530, grad_norm=4.380, loss_scale=1.000, learning_rate=1.085e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 19:54:57,695 (deepspeed_trainer:228) INFO: 14epoch:train:9001-9100batch: iter_time=1.093e-04, loss_ctc=77.181, loss_att=61.538, acc=0.694, loss=66.197, grad_norm=3.708, loss_scale=1.000, learning_rate=1.085e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 19:55:33,617 (deepspeed_trainer:228) INFO: 14epoch:train:9101-9200batch: iter_time=1.082e-04, loss_ctc=75.773, loss_att=67.519, acc=0.678, loss=70.001, grad_norm=3.909, loss_scale=1.000, learning_rate=1.084e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 19:56:09,363 (deepspeed_trainer:228) INFO: 14epoch:train:9201-9300batch: iter_time=1.090e-04, loss_ctc=74.108, loss_att=53.279, acc=0.686, loss=59.520, grad_norm=4.069, loss_scale=1.000, learning_rate=1.084e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-05 19:56:38,666 (multiple_iter_factory:32) INFO: Building 5th iter-factory... [cnode7-012:0/16] 2024-12-05 19:57:04,711 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-05 19:57:20,842 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.0", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.0", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.0", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.0", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-05 19:57:20,842 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.0, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-05 19:57:20,845 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-05 19:57:48,684] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:57:48,910] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:57:48,930] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:57:48,972] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:57:49,053] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:57:49,462] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:57:49,564] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:57:49,786] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:57:49,856] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:57:50,144] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:57:50,396] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:57:50,533] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:57:50,715] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:57:50,956] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:57:51,104] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:57:52,038] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:58:36,824] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:58:37,493] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:58:37,691] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:58:37,831] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:58:37,943] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:58:38,127] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:58:38,820] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:58:38,839] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:58:39,130] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:58:39,301] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:58:39,463] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:58:39,761] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:58:39,836] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:58:39,838] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:58:39,916] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:58:40,015] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:59:26,486] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:59:26,527] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:59:26,618] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:59:26,796] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:59:27,327] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:59:27,645] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:59:27,599] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:59:27,665] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:59:27,940] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:59:28,107] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:59:28,474] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:59:28,794] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:59:29,185] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:59:29,413] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:59:29,460] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 19:59:29,415] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:00:14,520] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:00:15,116] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:00:15,385] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:00:15,511] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:00:15,583] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:00:15,589] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:00:16,429] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:00:16,456] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:00:16,577] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:00:16,845] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:00:17,048] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:00:17,520] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:00:17,557] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:00:17,697] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:00:17,725] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:00:20,934] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-05 20:00:55,823 (deepspeed_trainer:228) INFO: 14epoch:train:9301-9400batch: iter_time=2.464, loss_ctc=74.649, loss_att=58.555, acc=0.672, loss=63.372, grad_norm=4.091, loss_scale=1.000, learning_rate=1.084e-04, step_time=0.400 [cnode7-012:0/16] 2024-12-05 20:01:31,908 (deepspeed_trainer:228) INFO: 14epoch:train:9401-9500batch: iter_time=1.064e-04, loss_ctc=87.347, loss_att=67.082, acc=0.676, loss=73.154, grad_norm=4.622, loss_scale=1.000, learning_rate=1.083e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 20:02:07,930 (deepspeed_trainer:228) INFO: 14epoch:train:9501-9600batch: iter_time=1.101e-04, loss_ctc=76.566, loss_att=61.223, acc=0.675, loss=65.833, grad_norm=3.952, loss_scale=1.000, learning_rate=1.083e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 20:02:43,731 (deepspeed_trainer:228) INFO: 14epoch:train:9601-9700batch: iter_time=1.089e-04, loss_ctc=77.344, loss_att=59.115, acc=0.693, loss=64.562, grad_norm=3.967, loss_scale=1.000, learning_rate=1.083e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 20:03:19,775 (deepspeed_trainer:228) INFO: 14epoch:train:9701-9800batch: iter_time=1.095e-04, loss_ctc=83.170, loss_att=70.971, acc=0.650, loss=74.636, grad_norm=4.271, loss_scale=1.000, learning_rate=1.083e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 20:03:55,482 (deepspeed_trainer:228) INFO: 14epoch:train:9801-9900batch: iter_time=1.120e-04, loss_ctc=80.223, loss_att=60.709, acc=0.694, loss=66.567, grad_norm=3.938, loss_scale=1.000, learning_rate=1.082e-04, step_time=0.356 [2024-12-05 20:04:31,171] [INFO] [logging.py:129:log_dist] [Rank 0] step=205000, skipped=0, lr=[np.float64(0.00010820009225774967)], mom=[[0.9, 0.98]] [2024-12-05 20:04:31,172] [INFO] [timer.py:264:stop] epoch=0/micro_step=25000/global_step=25000, RunningAvgSamplesPerSec=45.158208955889855, CurrSamplesPerSec=46.094285779622766, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 20:04:31,174 (deepspeed_trainer:228) INFO: 14epoch:train:9901-10000batch: iter_time=1.100e-04, loss_ctc=74.010, loss_att=57.108, acc=0.691, loss=62.175, grad_norm=4.165, loss_scale=1.000, learning_rate=1.082e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-05 20:05:07,464 (deepspeed_trainer:228) INFO: 14epoch:train:10001-10100batch: iter_time=1.127e-04, loss_ctc=90.137, loss_att=72.399, acc=0.664, loss=77.743, grad_norm=4.536, loss_scale=1.000, learning_rate=1.082e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-05 20:05:42,866 (deepspeed_trainer:228) INFO: 14epoch:train:10101-10200batch: iter_time=1.122e-04, loss_ctc=74.493, loss_att=55.548, acc=0.697, loss=61.239, grad_norm=3.941, loss_scale=1.000, learning_rate=1.082e-04, step_time=0.354 [cnode7-012:0/16] 2024-12-05 20:06:18,918 (deepspeed_trainer:228) INFO: 14epoch:train:10201-10300batch: iter_time=1.120e-04, loss_ctc=81.764, loss_att=58.359, acc=0.681, loss=65.362, grad_norm=4.597, loss_scale=1.000, learning_rate=1.081e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 20:06:55,007 (deepspeed_trainer:228) INFO: 14epoch:train:10301-10400batch: iter_time=1.164e-04, loss_ctc=69.608, loss_att=52.718, acc=0.692, loss=57.787, grad_norm=3.938, loss_scale=1.000, learning_rate=1.081e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 20:07:30,618 (deepspeed_trainer:228) INFO: 14epoch:train:10401-10500batch: iter_time=1.093e-04, loss_ctc=76.787, loss_att=64.187, acc=0.672, loss=67.975, grad_norm=4.109, loss_scale=1.000, learning_rate=1.081e-04, step_time=0.356 [cnode7-012:0/16] 2024-12-05 20:08:06,211 (deepspeed_trainer:228) INFO: 14epoch:train:10501-10600batch: iter_time=1.104e-04, loss_ctc=79.388, loss_att=61.562, acc=0.683, loss=66.936, grad_norm=4.289, loss_scale=1.000, learning_rate=1.081e-04, step_time=0.356 [cnode7-012:0/16] 2024-12-05 20:08:42,568 (deepspeed_trainer:228) INFO: 14epoch:train:10601-10700batch: iter_time=1.100e-04, loss_ctc=82.820, loss_att=61.689, acc=0.686, loss=68.021, grad_norm=4.457, loss_scale=1.000, learning_rate=1.080e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-05 20:09:18,532 (deepspeed_trainer:228) INFO: 14epoch:train:10701-10800batch: iter_time=1.076e-04, loss_ctc=80.561, loss_att=60.123, acc=0.695, loss=66.255, grad_norm=4.633, loss_scale=1.000, learning_rate=1.080e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 20:09:54,408 (deepspeed_trainer:228) INFO: 14epoch:train:10801-10900batch: iter_time=1.103e-04, loss_ctc=75.314, loss_att=56.530, acc=0.695, loss=62.185, grad_norm=4.305, loss_scale=1.000, learning_rate=1.080e-04, step_time=0.358 [2024-12-05 20:10:30,379] [INFO] [logging.py:129:log_dist] [Rank 0] step=206000, skipped=0, lr=[np.float64(0.00010793715246111136)], mom=[[0.9, 0.98]] [2024-12-05 20:10:30,380] [INFO] [timer.py:264:stop] epoch=0/micro_step=26000/global_step=26000, RunningAvgSamplesPerSec=45.1840514925116, CurrSamplesPerSec=44.85676157141832, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 20:10:30,381 (deepspeed_trainer:228) INFO: 14epoch:train:10901-11000batch: iter_time=1.102e-04, loss_ctc=76.881, loss_att=58.723, acc=0.688, loss=64.194, grad_norm=3.831, loss_scale=1.000, learning_rate=1.080e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 20:11:06,415 (deepspeed_trainer:228) INFO: 14epoch:train:11001-11100batch: iter_time=1.116e-04, loss_ctc=75.121, loss_att=65.363, acc=0.680, loss=68.319, grad_norm=3.965, loss_scale=1.000, learning_rate=1.079e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 20:11:42,196 (deepspeed_trainer:228) INFO: 14epoch:train:11101-11200batch: iter_time=1.115e-04, loss_ctc=75.605, loss_att=54.656, acc=0.688, loss=60.916, grad_norm=4.158, loss_scale=1.000, learning_rate=1.079e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-05 20:12:02,655 (multiple_iter_factory:32) INFO: Building 6th iter-factory... [cnode7-012:0/16] 2024-12-05 20:12:29,073 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-05 20:12:45,809 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.4", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.4", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.4", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.4", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-05 20:12:45,810 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.4, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-05 20:12:45,812 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-05 20:13:12,566] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:13:12,629] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:13:12,741] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:13:12,841] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:13:13,048] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:13:13,247] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:13:13,631] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:13:13,697] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:13:13,779] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:13:14,025] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:13:14,278] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:13:14,568] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:13:14,616] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:13:14,689] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:13:15,275] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:13:15,802] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:14:00,741] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:14:00,791] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:14:01,030] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:14:01,226] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:14:01,389] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:14:02,217] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:14:02,211] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:14:02,638] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:14:02,696] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:14:02,847] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:14:03,177] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:14:03,291] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:14:03,674] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:14:03,643] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:14:03,944] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:14:04,753] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:14:47,608] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:14:47,704] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:14:48,442] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:14:48,642] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:14:50,630] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:14:50,817] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:14:50,924] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:14:51,080] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:14:51,433] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:14:51,850] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:14:51,928] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:14:51,981] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:14:52,119] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:14:52,241] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:14:52,532] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:14:55,430] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:15:34,282] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:15:34,280] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:15:35,366] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:15:38,244] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:15:38,726] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:15:39,500] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:15:40,002] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:15:40,028] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:15:39,997] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:15:40,351] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:15:40,509] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:15:40,655] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:15:40,894] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:15:43,191] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:15:45,519] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:15:45,842] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-05 20:16:30,108 (deepspeed_trainer:228) INFO: 14epoch:train:11201-11300batch: iter_time=2.512, loss_ctc=77.001, loss_att=62.046, acc=0.668, loss=66.545, grad_norm=4.285, loss_scale=1.000, learning_rate=1.079e-04, step_time=0.367 [cnode7-012:0/16] 2024-12-05 20:17:06,263 (deepspeed_trainer:228) INFO: 14epoch:train:11301-11400batch: iter_time=1.114e-04, loss_ctc=84.199, loss_att=63.707, acc=0.682, loss=69.840, grad_norm=4.789, loss_scale=1.000, learning_rate=1.078e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 20:17:42,484 (deepspeed_trainer:228) INFO: 14epoch:train:11401-11500batch: iter_time=1.101e-04, loss_ctc=79.188, loss_att=62.652, acc=0.675, loss=67.626, grad_norm=4.288, loss_scale=1.000, learning_rate=1.078e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 20:18:18,712 (deepspeed_trainer:228) INFO: 14epoch:train:11501-11600batch: iter_time=1.092e-04, loss_ctc=75.249, loss_att=61.300, acc=0.690, loss=65.484, grad_norm=4.107, loss_scale=1.000, learning_rate=1.078e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 20:18:55,405 (deepspeed_trainer:228) INFO: 14epoch:train:11601-11700batch: iter_time=1.093e-04, loss_ctc=83.999, loss_att=69.001, acc=0.661, loss=73.534, grad_norm=4.402, loss_scale=1.000, learning_rate=1.078e-04, step_time=0.366 [cnode7-012:0/16] 2024-12-05 20:19:31,738 (deepspeed_trainer:228) INFO: 14epoch:train:11701-11800batch: iter_time=1.089e-04, loss_ctc=77.268, loss_att=61.186, acc=0.688, loss=66.008, grad_norm=4.078, loss_scale=1.000, learning_rate=1.077e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-05 20:20:07,608 (deepspeed_trainer:228) INFO: 14epoch:train:11801-11900batch: iter_time=1.087e-04, loss_ctc=73.345, loss_att=55.948, acc=0.694, loss=61.186, grad_norm=4.096, loss_scale=1.000, learning_rate=1.077e-04, step_time=0.358 [2024-12-05 20:20:43,791] [INFO] [logging.py:129:log_dist] [Rank 0] step=207000, skipped=0, lr=[np.float64(0.00010767612032470005)], mom=[[0.9, 0.98]] [2024-12-05 20:20:43,791] [INFO] [timer.py:264:stop] epoch=0/micro_step=27000/global_step=27000, RunningAvgSamplesPerSec=45.192367193210735, CurrSamplesPerSec=47.69753287190004, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 20:20:43,793 (deepspeed_trainer:228) INFO: 14epoch:train:11901-12000batch: iter_time=1.091e-04, loss_ctc=91.588, loss_att=69.516, acc=0.666, loss=76.106, grad_norm=4.822, loss_scale=1.000, learning_rate=1.077e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 20:21:19,510 (deepspeed_trainer:228) INFO: 14epoch:train:12001-12100batch: iter_time=1.102e-04, loss_ctc=72.171, loss_att=54.626, acc=0.696, loss=59.911, grad_norm=4.016, loss_scale=1.000, learning_rate=1.077e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-05 20:21:55,295 (deepspeed_trainer:228) INFO: 14epoch:train:12101-12200batch: iter_time=1.096e-04, loss_ctc=79.967, loss_att=56.974, acc=0.688, loss=63.825, grad_norm=4.581, loss_scale=1.000, learning_rate=1.076e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 20:22:31,176 (deepspeed_trainer:228) INFO: 14epoch:train:12201-12300batch: iter_time=1.089e-04, loss_ctc=74.571, loss_att=58.456, acc=0.684, loss=63.316, grad_norm=3.844, loss_scale=1.000, learning_rate=1.076e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 20:23:07,220 (deepspeed_trainer:228) INFO: 14epoch:train:12301-12400batch: iter_time=1.106e-04, loss_ctc=75.207, loss_att=60.771, acc=0.677, loss=65.083, grad_norm=4.228, loss_scale=1.000, learning_rate=1.076e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 20:23:43,296 (deepspeed_trainer:228) INFO: 14epoch:train:12401-12500batch: iter_time=1.091e-04, loss_ctc=82.159, loss_att=62.985, acc=0.684, loss=68.741, grad_norm=4.389, loss_scale=1.000, learning_rate=1.076e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 20:24:19,298 (deepspeed_trainer:228) INFO: 14epoch:train:12501-12600batch: iter_time=1.081e-04, loss_ctc=80.084, loss_att=61.207, acc=0.689, loss=66.844, grad_norm=4.473, loss_scale=1.000, learning_rate=1.075e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 20:24:55,578 (deepspeed_trainer:228) INFO: 14epoch:train:12601-12700batch: iter_time=1.102e-04, loss_ctc=78.429, loss_att=57.500, acc=0.698, loss=63.785, grad_norm=3.749, loss_scale=1.000, learning_rate=1.075e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-05 20:25:31,795 (deepspeed_trainer:228) INFO: 14epoch:train:12701-12800batch: iter_time=1.103e-04, loss_ctc=80.697, loss_att=60.455, acc=0.693, loss=66.516, grad_norm=4.297, loss_scale=1.000, learning_rate=1.075e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 20:26:08,106 (deepspeed_trainer:228) INFO: 14epoch:train:12801-12900batch: iter_time=1.078e-04, loss_ctc=72.730, loss_att=62.083, acc=0.686, loss=65.312, grad_norm=4.332, loss_scale=1.000, learning_rate=1.075e-04, step_time=0.363 [2024-12-05 20:26:44,223] [INFO] [logging.py:129:log_dist] [Rank 0] step=208000, skipped=0, lr=[np.float64(0.00010741697289234815)], mom=[[0.9, 0.98]] [2024-12-05 20:26:44,225] [INFO] [timer.py:264:stop] epoch=0/micro_step=28000/global_step=28000, RunningAvgSamplesPerSec=45.207646224986945, CurrSamplesPerSec=46.148456729097724, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 20:26:44,228 (deepspeed_trainer:228) INFO: 14epoch:train:12901-13000batch: iter_time=1.119e-04, loss_ctc=74.781, loss_att=56.136, acc=0.692, loss=61.720, grad_norm=4.240, loss_scale=1.000, learning_rate=1.074e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 20:27:20,286 (deepspeed_trainer:228) INFO: 14epoch:train:13001-13100batch: iter_time=1.088e-04, loss_ctc=75.381, loss_att=58.451, acc=0.673, loss=63.535, grad_norm=4.726, loss_scale=1.000, learning_rate=1.074e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 20:27:31,832 (multiple_iter_factory:32) INFO: Building 7th iter-factory... [cnode7-012:0/16] 2024-12-05 20:27:58,463 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-05 20:28:14,664 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.5", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.5", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.5", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.5", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-05 20:28:14,664 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.5, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-05 20:28:14,666 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-05 20:28:41,431] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:28:41,415] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:28:41,678] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:28:41,755] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:28:42,607] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:28:42,708] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:28:42,842] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:28:43,204] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:28:43,542] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:28:43,569] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:28:44,133] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:28:44,172] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:28:44,575] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:28:44,735] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:28:44,828] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:28:45,248] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:29:29,395] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:29:29,551] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:29:29,914] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:29:30,458] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:29:31,096] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:29:31,850] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:29:32,557] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:29:32,580] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:29:32,625] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:29:32,841] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:29:33,050] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:29:33,122] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:29:33,171] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:29:33,718] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:29:33,817] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:29:33,869] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:30:17,316] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:30:17,710] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:30:17,969] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:30:18,180] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:30:18,748] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:30:20,580] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:30:20,973] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:30:21,099] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:30:21,119] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:30:21,355] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:30:22,209] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:30:22,307] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:30:22,635] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:30:24,006] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:30:25,174] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:30:27,572] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:31:05,342] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:31:05,945] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:31:06,083] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:31:06,212] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:31:08,466] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:31:09,470] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:31:09,525] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:31:11,133] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:31:11,196] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:31:11,553] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:31:11,753] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:31:11,882] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:31:11,923] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:31:14,271] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:31:16,668] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:31:18,722] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-05 20:32:12,260 (deepspeed_trainer:228) INFO: 14epoch:train:13101-13200batch: iter_time=2.460, loss_ctc=78.829, loss_att=60.708, acc=0.680, loss=66.146, grad_norm=4.344, loss_scale=1.000, learning_rate=1.074e-04, step_time=0.459 [cnode7-012:0/16] 2024-12-05 20:32:48,657 (deepspeed_trainer:228) INFO: 14epoch:train:13201-13300batch: iter_time=1.107e-04, loss_ctc=83.724, loss_att=62.853, acc=0.670, loss=69.126, grad_norm=4.773, loss_scale=1.000, learning_rate=1.074e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-05 20:33:25,071 (deepspeed_trainer:228) INFO: 14epoch:train:13301-13400batch: iter_time=1.116e-04, loss_ctc=75.564, loss_att=59.312, acc=0.681, loss=64.197, grad_norm=3.879, loss_scale=1.000, learning_rate=1.073e-04, step_time=0.364 [cnode7-012:0/16] 2024-12-05 20:34:01,705 (deepspeed_trainer:228) INFO: 14epoch:train:13401-13500batch: iter_time=1.162e-04, loss_ctc=79.257, loss_att=66.211, acc=0.670, loss=70.117, grad_norm=4.314, loss_scale=1.000, learning_rate=1.073e-04, step_time=0.366 [cnode7-012:0/16] 2024-12-05 20:34:38,237 (deepspeed_trainer:228) INFO: 14epoch:train:13501-13600batch: iter_time=1.113e-04, loss_ctc=86.091, loss_att=68.249, acc=0.665, loss=73.590, grad_norm=4.498, loss_scale=1.000, learning_rate=1.073e-04, step_time=0.365 [cnode7-012:0/16] 2024-12-05 20:35:14,380 (deepspeed_trainer:228) INFO: 14epoch:train:13601-13700batch: iter_time=1.114e-04, loss_ctc=70.744, loss_att=54.981, acc=0.695, loss=59.727, grad_norm=3.751, loss_scale=1.000, learning_rate=1.072e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 20:35:50,560 (deepspeed_trainer:228) INFO: 14epoch:train:13701-13800batch: iter_time=1.111e-04, loss_ctc=77.255, loss_att=61.637, acc=0.682, loss=66.313, grad_norm=4.532, loss_scale=1.000, learning_rate=1.072e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 20:36:26,547 (deepspeed_trainer:228) INFO: 14epoch:train:13801-13900batch: iter_time=1.098e-04, loss_ctc=84.856, loss_att=61.918, acc=0.674, loss=68.804, grad_norm=5.050, loss_scale=1.000, learning_rate=1.072e-04, step_time=0.359 [2024-12-05 20:37:02,230] [INFO] [logging.py:129:log_dist] [Rank 0] step=209000, skipped=0, lr=[np.float64(0.00010715968759278164)], mom=[[0.9, 0.98]] [2024-12-05 20:37:02,231] [INFO] [timer.py:264:stop] epoch=0/micro_step=29000/global_step=29000, RunningAvgSamplesPerSec=45.17225531238559, CurrSamplesPerSec=46.61832820104253, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 20:37:02,233 (deepspeed_trainer:228) INFO: 14epoch:train:13901-14000batch: iter_time=1.118e-04, loss_ctc=73.921, loss_att=54.789, acc=0.689, loss=60.538, grad_norm=3.943, loss_scale=1.000, learning_rate=1.072e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-05 20:37:38,042 (deepspeed_trainer:228) INFO: 14epoch:train:14001-14100batch: iter_time=1.091e-04, loss_ctc=78.452, loss_att=57.858, acc=0.683, loss=64.024, grad_norm=4.434, loss_scale=1.000, learning_rate=1.071e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 20:38:14,296 (deepspeed_trainer:228) INFO: 14epoch:train:14101-14200batch: iter_time=1.112e-04, loss_ctc=73.517, loss_att=59.579, acc=0.675, loss=63.753, grad_norm=4.184, loss_scale=1.000, learning_rate=1.071e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 20:38:50,287 (deepspeed_trainer:228) INFO: 14epoch:train:14201-14300batch: iter_time=1.109e-04, loss_ctc=76.166, loss_att=61.999, acc=0.673, loss=66.245, grad_norm=4.024, loss_scale=1.000, learning_rate=1.071e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 20:39:26,522 (deepspeed_trainer:228) INFO: 14epoch:train:14301-14400batch: iter_time=1.090e-04, loss_ctc=82.611, loss_att=61.866, acc=0.683, loss=68.101, grad_norm=4.423, loss_scale=1.000, learning_rate=1.071e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 20:40:02,851 (deepspeed_trainer:228) INFO: 14epoch:train:14401-14500batch: iter_time=1.132e-04, loss_ctc=80.085, loss_att=58.790, acc=0.692, loss=65.206, grad_norm=4.159, loss_scale=1.000, learning_rate=1.070e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-05 20:40:38,951 (deepspeed_trainer:228) INFO: 14epoch:train:14501-14600batch: iter_time=1.114e-04, loss_ctc=78.303, loss_att=57.463, acc=0.689, loss=63.710, grad_norm=3.871, loss_scale=1.000, learning_rate=1.070e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 20:41:14,855 (deepspeed_trainer:228) INFO: 14epoch:train:14601-14700batch: iter_time=1.113e-04, loss_ctc=76.816, loss_att=57.562, acc=0.694, loss=63.357, grad_norm=4.021, loss_scale=1.000, learning_rate=1.070e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 20:41:50,854 (deepspeed_trainer:228) INFO: 14epoch:train:14701-14800batch: iter_time=1.092e-04, loss_ctc=73.794, loss_att=65.779, acc=0.676, loss=68.201, grad_norm=3.834, loss_scale=1.000, learning_rate=1.070e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 20:42:26,954 (deepspeed_trainer:228) INFO: 14epoch:train:14801-14900batch: iter_time=1.086e-04, loss_ctc=76.864, loss_att=54.974, acc=0.685, loss=61.513, grad_norm=4.007, loss_scale=1.000, learning_rate=1.069e-04, step_time=0.361 [2024-12-05 20:43:02,804] [INFO] [logging.py:129:log_dist] [Rank 0] step=210000, skipped=0, lr=[np.float64(0.0001069042422313627)], mom=[[0.9, 0.98]] [2024-12-05 20:43:02,805] [INFO] [timer.py:264:stop] epoch=0/micro_step=30000/global_step=30000, RunningAvgSamplesPerSec=45.18771133797291, CurrSamplesPerSec=47.036301388375016, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 20:43:02,807 (deepspeed_trainer:228) INFO: 14epoch:train:14901-15000batch: iter_time=1.077e-04, loss_ctc=70.669, loss_att=55.006, acc=0.675, loss=59.716, grad_norm=4.553, loss_scale=1.000, learning_rate=1.069e-04, step_time=0.358 [2024-12-05 20:43:17,137] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:17,635] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:17,801] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:18,556] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:18,618] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:18,699] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:18,793] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:18,831] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:18,873] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:18,941] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:18,951] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:18,973] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:19,011] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:19,015] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:19,070] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:19,087] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:32,720] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:33,575] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:33,620] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:34,202] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:34,444] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:34,552] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:35,278] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:35,283] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:35,528] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:35,767] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:35,996] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:36,030] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:36,095] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:36,147] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:36,187] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:36,225] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:47,613] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:49,837] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:49,876] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:50,243] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:50,620] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:50,808] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:51,107] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:51,190] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:51,256] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:51,471] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:51,901] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:52,098] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:52,319] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:52,436] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:52,529] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:43:52,579] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:44:02,319] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:44:05,003] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:44:05,392] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:44:05,720] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:44:06,307] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:44:06,552] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:44:06,927] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:44:07,139] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:44:07,201] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:44:07,283] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:44:07,591] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:44:07,675] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:44:07,947] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:44:08,137] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:44:08,317] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:44:08,488] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:44:21,221] [INFO] [logging.py:129:log_dist] [Rank 0] [Torch] Checkpoint 14 is about to be saved! [2024-12-05 20:44:21,256] [INFO] [logging.py:129:log_dist] [Rank 0] Saving model checkpoint: exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/mp_rank_00_model_states.pt [2024-12-05 20:44:21,256] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/mp_rank_00_model_states.pt... [2024-12-05 20:44:23,098] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/mp_rank_00_model_states.pt. [2024-12-05 20:44:23,255] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_14_mp_rank_00_optim_states.pt... [2024-12-05 20:44:23,256] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_15_mp_rank_00_optim_states.pt... [2024-12-05 20:44:23,257] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_12_mp_rank_00_optim_states.pt... [2024-12-05 20:44:23,256] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt... [2024-12-05 20:44:23,257] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt... [2024-12-05 20:44:23,258] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt... [2024-12-05 20:44:23,259] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_9_mp_rank_00_optim_states.pt... [2024-12-05 20:44:23,260] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt... [2024-12-05 20:44:23,261] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_10_mp_rank_00_optim_states.pt... [2024-12-05 20:44:23,261] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt... [2024-12-05 20:44:23,262] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_11_mp_rank_00_optim_states.pt... [2024-12-05 20:44:23,262] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_13_mp_rank_00_optim_states.pt... [2024-12-05 20:44:23,262] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt... [2024-12-05 20:44:23,262] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt... [2024-12-05 20:44:23,263] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_8_mp_rank_00_optim_states.pt... [2024-12-05 20:44:23,263] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt... [2024-12-05 20:44:23,851] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt. [2024-12-05 20:44:23,852] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt [2024-12-05 20:44:23,852] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 14 is ready now! [2024-12-05 20:44:23,974] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt. [2024-12-05 20:44:23,974] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt [2024-12-05 20:44:23,974] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 14 is ready now! [2024-12-05 20:44:24,030] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt. [2024-12-05 20:44:24,030] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt [2024-12-05 20:44:24,030] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 14 is ready now! [2024-12-05 20:44:24,069] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt. [2024-12-05 20:44:24,069] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt [2024-12-05 20:44:24,069] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 14 is ready now! [2024-12-05 20:44:24,078] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt. [2024-12-05 20:44:24,079] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt [2024-12-05 20:44:24,079] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 14 is ready now! [2024-12-05 20:44:24,404] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_9_mp_rank_00_optim_states.pt. [2024-12-05 20:44:24,404] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_9_mp_rank_00_optim_states.pt [2024-12-05 20:44:24,404] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 14 is ready now! [2024-12-05 20:44:24,404] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_11_mp_rank_00_optim_states.pt. [2024-12-05 20:44:24,404] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_11_mp_rank_00_optim_states.pt [2024-12-05 20:44:24,404] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 14 is ready now! [2024-12-05 20:44:24,404] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_10_mp_rank_00_optim_states.pt. [2024-12-05 20:44:24,404] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_13_mp_rank_00_optim_states.pt. [2024-12-05 20:44:24,404] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_10_mp_rank_00_optim_states.pt [2024-12-05 20:44:24,404] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_14_mp_rank_00_optim_states.pt. [2024-12-05 20:44:24,404] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_13_mp_rank_00_optim_states.pt [2024-12-05 20:44:24,404] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 14 is ready now! [2024-12-05 20:44:24,404] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_14_mp_rank_00_optim_states.pt [2024-12-05 20:44:24,404] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 14 is ready now! [2024-12-05 20:44:24,404] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 14 is ready now! [2024-12-05 20:44:24,779] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt. [2024-12-05 20:44:24,779] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt [2024-12-05 20:44:24,779] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 14 is ready now! [2024-12-05 20:44:24,846] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt. [2024-12-05 20:44:24,846] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt. [2024-12-05 20:44:24,846] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt [2024-12-05 20:44:24,846] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 14 is ready now! [2024-12-05 20:44:24,856] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt [2024-12-05 20:44:24,856] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 14 is ready now! [2024-12-05 20:44:24,868] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_15_mp_rank_00_optim_states.pt. [2024-12-05 20:44:24,868] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_15_mp_rank_00_optim_states.pt [2024-12-05 20:44:24,868] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 14 is ready now! [2024-12-05 20:44:25,341] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_12_mp_rank_00_optim_states.pt. [2024-12-05 20:44:25,341] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_12_mp_rank_00_optim_states.pt [2024-12-05 20:44:25,341] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 14 is ready now! [2024-12-05 20:44:25,346] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_8_mp_rank_00_optim_states.pt. [2024-12-05 20:44:25,346] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_14/14/bf16_zero_pp_rank_8_mp_rank_00_optim_states.pt [2024-12-05 20:44:25,346] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 14 is ready now! [cnode7-012:0/16] 2024-12-05 20:44:25,354 (deepspeed_trainer:158) INFO: 14epoch results: [train] iter_time=0.131, loss_ctc=79.130, loss_att=61.196, acc=0.679, loss=66.578, grad_norm=4.261, loss_scale=1.000, learning_rate=1.089e-04, step_time=0.363, time=2 hours, 3 minutes and 51.84 seconds, total_count=210014, gpu_max_cached_mem_GB=36.545, [valid] loss_ctc=6.406, cer_ctc=0.194, loss_att=7.625, acc=0.795, cer=0.327, wer=1.000, loss=7.250, time=1 minute and 9.61 seconds, total_count=14, gpu_max_cached_mem_GB=36.545 [cnode7-012:0/16] 2024-12-05 20:44:26,408 (multiple_iter_factory:32) INFO: Building 0th iter-factory... [cnode7-012:0/16] 2024-12-05 20:44:52,493 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-05 20:45:07,854 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.1", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.1", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.1", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.1", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-05 20:45:07,854 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.1, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-05 20:45:07,857 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-05 20:45:30,402] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:45:30,632] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:45:31,213] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:45:31,752] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:45:32,108] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:45:32,213] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:45:32,451] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:45:32,493] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:45:32,580] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:45:32,667] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:45:33,216] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:45:33,275] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:45:33,313] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:45:33,384] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:45:33,388] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:45:33,797] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:46:16,650] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:46:18,196] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:46:18,873] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:46:19,037] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:46:19,168] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:46:19,571] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:46:19,951] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:46:20,446] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:46:20,645] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:46:20,690] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:46:20,986] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:46:21,096] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:46:21,098] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:46:21,170] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:46:21,212] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:46:21,503] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:47:02,495] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:47:04,184] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:47:06,725] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:47:07,012] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:47:07,162] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:47:07,234] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:47:08,024] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:47:08,249] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:47:08,287] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:47:08,733] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:47:08,951] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:47:09,121] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:47:09,228] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:47:09,270] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:47:10,407] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:47:11,153] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:47:50,282] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:47:51,022] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:47:54,246] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:47:54,464] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:47:54,771] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:47:55,218] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:47:55,804] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:47:56,381] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:47:56,500] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:47:56,498] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:47:56,982] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:47:57,025] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:47:57,504] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:47:57,810] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:47:58,193] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 20:47:59,700] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-05 20:49:01,539 (deepspeed_trainer:228) INFO: 15epoch:train:1-100batch: iter_time=2.301, loss_ctc=73.671, loss_att=58.483, acc=0.658, loss=63.038, grad_norm=4.465, loss_scale=1.000, learning_rate=1.069e-04, step_time=0.451 [cnode7-012:0/16] 2024-12-05 20:49:37,785 (deepspeed_trainer:228) INFO: 15epoch:train:101-200batch: iter_time=1.169e-04, loss_ctc=86.591, loss_att=60.922, acc=0.685, loss=68.638, grad_norm=4.532, loss_scale=1.000, learning_rate=1.069e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 20:50:13,708 (deepspeed_trainer:228) INFO: 15epoch:train:201-300batch: iter_time=1.118e-04, loss_ctc=83.249, loss_att=72.713, acc=0.650, loss=75.872, grad_norm=4.468, loss_scale=1.000, learning_rate=1.068e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 20:50:49,784 (deepspeed_trainer:228) INFO: 15epoch:train:301-400batch: iter_time=1.107e-04, loss_ctc=80.623, loss_att=63.151, acc=0.673, loss=68.385, grad_norm=4.440, loss_scale=1.000, learning_rate=1.068e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 20:51:25,629 (deepspeed_trainer:228) INFO: 15epoch:train:401-500batch: iter_time=1.109e-04, loss_ctc=79.704, loss_att=63.590, acc=0.674, loss=68.431, grad_norm=4.243, loss_scale=1.000, learning_rate=1.068e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 20:52:01,440 (deepspeed_trainer:228) INFO: 15epoch:train:501-600batch: iter_time=1.107e-04, loss_ctc=77.464, loss_att=60.546, acc=0.673, loss=65.661, grad_norm=4.187, loss_scale=1.000, learning_rate=1.068e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 20:52:37,030 (deepspeed_trainer:228) INFO: 15epoch:train:601-700batch: iter_time=1.114e-04, loss_ctc=71.476, loss_att=56.988, acc=0.667, loss=61.356, grad_norm=4.085, loss_scale=1.000, learning_rate=1.067e-04, step_time=0.355 [cnode7-012:0/16] 2024-12-05 20:53:13,084 (deepspeed_trainer:228) INFO: 15epoch:train:701-800batch: iter_time=1.131e-04, loss_ctc=74.081, loss_att=59.437, acc=0.681, loss=63.805, grad_norm=3.749, loss_scale=1.000, learning_rate=1.067e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 20:53:49,389 (deepspeed_trainer:228) INFO: 15epoch:train:801-900batch: iter_time=1.125e-04, loss_ctc=83.640, loss_att=73.548, acc=0.663, loss=76.596, grad_norm=4.090, loss_scale=1.000, learning_rate=1.067e-04, step_time=0.363 [2024-12-05 20:54:25,614] [INFO] [logging.py:129:log_dist] [Rank 0] step=211000, skipped=0, lr=[np.float64(0.00010665061498204748)], mom=[[0.9, 0.98]] [2024-12-05 20:54:25,615] [INFO] [timer.py:264:stop] epoch=0/micro_step=31000/global_step=31000, RunningAvgSamplesPerSec=45.16790411591155, CurrSamplesPerSec=44.489443508198086, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 20:54:25,617 (deepspeed_trainer:228) INFO: 15epoch:train:901-1000batch: iter_time=1.106e-04, loss_ctc=77.126, loss_att=65.910, acc=0.678, loss=69.262, grad_norm=3.919, loss_scale=1.000, learning_rate=1.067e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 20:55:01,694 (deepspeed_trainer:228) INFO: 15epoch:train:1001-1100batch: iter_time=1.087e-04, loss_ctc=83.202, loss_att=58.716, acc=0.686, loss=66.061, grad_norm=4.283, loss_scale=1.000, learning_rate=1.066e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 20:55:37,462 (deepspeed_trainer:228) INFO: 15epoch:train:1101-1200batch: iter_time=1.119e-04, loss_ctc=91.501, loss_att=67.045, acc=0.680, loss=74.374, grad_norm=5.026, loss_scale=1.000, learning_rate=1.066e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-05 20:56:13,570 (deepspeed_trainer:228) INFO: 15epoch:train:1201-1300batch: iter_time=1.108e-04, loss_ctc=84.992, loss_att=62.101, acc=0.698, loss=68.996, grad_norm=3.908, loss_scale=1.000, learning_rate=1.066e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 20:56:50,008 (deepspeed_trainer:228) INFO: 15epoch:train:1301-1400batch: iter_time=1.114e-04, loss_ctc=80.739, loss_att=59.574, acc=0.686, loss=65.924, grad_norm=4.324, loss_scale=1.000, learning_rate=1.066e-04, step_time=0.364 [cnode7-012:0/16] 2024-12-05 20:57:26,340 (deepspeed_trainer:228) INFO: 15epoch:train:1401-1500batch: iter_time=1.111e-04, loss_ctc=81.398, loss_att=64.894, acc=0.660, loss=69.842, grad_norm=4.438, loss_scale=1.000, learning_rate=1.065e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-05 20:58:02,359 (deepspeed_trainer:228) INFO: 15epoch:train:1501-1600batch: iter_time=1.098e-04, loss_ctc=73.216, loss_att=61.803, acc=0.677, loss=65.221, grad_norm=3.833, loss_scale=1.000, learning_rate=1.065e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 20:58:38,314 (deepspeed_trainer:228) INFO: 15epoch:train:1601-1700batch: iter_time=1.116e-04, loss_ctc=83.376, loss_att=64.447, acc=0.678, loss=70.141, grad_norm=4.464, loss_scale=1.000, learning_rate=1.065e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 20:59:14,380 (deepspeed_trainer:228) INFO: 15epoch:train:1701-1800batch: iter_time=1.154e-04, loss_ctc=84.137, loss_att=53.403, acc=0.686, loss=62.611, grad_norm=4.309, loss_scale=1.000, learning_rate=1.065e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 20:59:44,383 (multiple_iter_factory:32) INFO: Building 1th iter-factory... [cnode7-012:0/16] 2024-12-05 21:00:10,507 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-05 21:00:26,705 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.3", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.3", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.3", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.3", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-05 21:00:26,705 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.3, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-05 21:00:26,707 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-05 21:00:55,578] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:00:55,732] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:00:55,960] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:00:56,083] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:00:56,167] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:00:56,206] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:00:56,283] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:00:56,386] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:00:56,514] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:00:57,271] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:00:57,445] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:00:57,622] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:00:57,879] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:00:58,283] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:00:58,340] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:01:04,341] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:01:44,511] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:01:44,918] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:01:45,481] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:01:45,671] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:01:45,853] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:01:46,080] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:01:46,189] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:01:46,982] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:01:47,109] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:01:47,093] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:01:47,179] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:01:47,300] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:01:47,430] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:01:48,011] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:01:48,904] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:01:51,645] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:02:33,026] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:02:33,175] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:02:33,464] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:02:33,776] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:02:34,050] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:02:34,372] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:02:35,087] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:02:35,159] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:02:35,846] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:02:35,890] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:02:35,966] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:02:36,392] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:02:36,429] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:02:38,872] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:02:39,992] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:02:40,632] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:03:20,471] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:03:20,837] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:03:21,655] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:03:21,798] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:03:21,860] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:03:22,938] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:03:23,170] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:03:23,305] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:03:23,784] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:03:24,409] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:03:24,672] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:03:24,858] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:03:25,015] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:03:26,326] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:03:30,440] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:03:30,833] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-05 21:04:05,847 (deepspeed_trainer:228) INFO: 15epoch:train:1801-1900batch: iter_time=2.453, loss_ctc=90.161, loss_att=65.029, acc=0.677, loss=72.541, grad_norm=4.854, loss_scale=1.000, learning_rate=1.064e-04, step_time=0.462 [2024-12-05 21:04:43,105] [INFO] [logging.py:129:log_dist] [Rank 0] step=212000, skipped=0, lr=[np.float64(0.00010639878437955302)], mom=[[0.9, 0.98]] [2024-12-05 21:04:43,105] [INFO] [timer.py:264:stop] epoch=0/micro_step=32000/global_step=32000, RunningAvgSamplesPerSec=45.135974521628256, CurrSamplesPerSec=43.683579622903416, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 21:04:43,107 (deepspeed_trainer:228) INFO: 15epoch:train:1901-2000batch: iter_time=1.139e-04, loss_ctc=82.106, loss_att=60.821, acc=0.667, loss=67.197, grad_norm=4.756, loss_scale=1.000, learning_rate=1.064e-04, step_time=0.372 [cnode7-012:0/16] 2024-12-05 21:05:19,653 (deepspeed_trainer:228) INFO: 15epoch:train:2001-2100batch: iter_time=1.195e-04, loss_ctc=81.935, loss_att=63.287, acc=0.680, loss=68.866, grad_norm=4.236, loss_scale=1.000, learning_rate=1.064e-04, step_time=0.365 [cnode7-012:0/16] 2024-12-05 21:05:55,446 (deepspeed_trainer:228) INFO: 15epoch:train:2101-2200batch: iter_time=1.127e-04, loss_ctc=77.014, loss_att=63.623, acc=0.664, loss=67.645, grad_norm=4.227, loss_scale=1.000, learning_rate=1.064e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 21:06:31,821 (deepspeed_trainer:228) INFO: 15epoch:train:2201-2300batch: iter_time=1.119e-04, loss_ctc=81.424, loss_att=65.073, acc=0.674, loss=70.010, grad_norm=4.394, loss_scale=1.000, learning_rate=1.063e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-05 21:07:08,157 (deepspeed_trainer:228) INFO: 15epoch:train:2301-2400batch: iter_time=1.116e-04, loss_ctc=86.183, loss_att=66.616, acc=0.676, loss=72.478, grad_norm=4.451, loss_scale=1.000, learning_rate=1.063e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-05 21:07:44,183 (deepspeed_trainer:228) INFO: 15epoch:train:2401-2500batch: iter_time=1.149e-04, loss_ctc=68.368, loss_att=52.513, acc=0.677, loss=57.273, grad_norm=4.507, loss_scale=1.000, learning_rate=1.063e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 21:08:19,913 (deepspeed_trainer:228) INFO: 15epoch:train:2501-2600batch: iter_time=1.234e-04, loss_ctc=71.908, loss_att=61.174, acc=0.675, loss=64.409, grad_norm=3.855, loss_scale=1.000, learning_rate=1.063e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-05 21:08:55,834 (deepspeed_trainer:228) INFO: 15epoch:train:2601-2700batch: iter_time=1.219e-04, loss_ctc=74.995, loss_att=56.959, acc=0.689, loss=62.350, grad_norm=4.337, loss_scale=1.000, learning_rate=1.062e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 21:09:32,859 (deepspeed_trainer:228) INFO: 15epoch:train:2701-2800batch: iter_time=1.127e-04, loss_ctc=78.109, loss_att=71.135, acc=0.670, loss=73.204, grad_norm=4.078, loss_scale=1.000, learning_rate=1.062e-04, step_time=0.370 [cnode7-012:0/16] 2024-12-05 21:10:09,527 (deepspeed_trainer:228) INFO: 15epoch:train:2801-2900batch: iter_time=1.141e-04, loss_ctc=81.492, loss_att=65.907, acc=0.681, loss=70.575, grad_norm=4.037, loss_scale=1.000, learning_rate=1.062e-04, step_time=0.366 [2024-12-05 21:10:45,647] [INFO] [logging.py:129:log_dist] [Rank 0] step=213000, skipped=0, lr=[np.float64(0.00010614872931172658)], mom=[[0.9, 0.98]] [2024-12-05 21:10:45,647] [INFO] [timer.py:264:stop] epoch=0/micro_step=33000/global_step=33000, RunningAvgSamplesPerSec=45.14410604281724, CurrSamplesPerSec=47.25343427180076, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 21:10:45,649 (deepspeed_trainer:228) INFO: 15epoch:train:2901-3000batch: iter_time=1.124e-04, loss_ctc=81.483, loss_att=57.628, acc=0.685, loss=64.789, grad_norm=4.281, loss_scale=1.000, learning_rate=1.062e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 21:11:22,281 (deepspeed_trainer:228) INFO: 15epoch:train:3001-3100batch: iter_time=1.154e-04, loss_ctc=85.288, loss_att=62.694, acc=0.694, loss=69.466, grad_norm=4.504, loss_scale=1.000, learning_rate=1.061e-04, step_time=0.366 [cnode7-012:0/16] 2024-12-05 21:11:58,540 (deepspeed_trainer:228) INFO: 15epoch:train:3101-3200batch: iter_time=1.133e-04, loss_ctc=88.192, loss_att=66.844, acc=0.695, loss=73.261, grad_norm=3.973, loss_scale=1.000, learning_rate=1.061e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 21:12:34,638 (deepspeed_trainer:228) INFO: 15epoch:train:3201-3300batch: iter_time=1.102e-04, loss_ctc=75.499, loss_att=54.704, acc=0.684, loss=60.979, grad_norm=3.861, loss_scale=1.000, learning_rate=1.061e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 21:13:11,063 (deepspeed_trainer:228) INFO: 15epoch:train:3301-3400batch: iter_time=1.121e-04, loss_ctc=81.571, loss_att=66.705, acc=0.664, loss=71.176, grad_norm=4.097, loss_scale=1.000, learning_rate=1.061e-04, step_time=0.364 [cnode7-012:0/16] 2024-12-05 21:13:47,332 (deepspeed_trainer:228) INFO: 15epoch:train:3401-3500batch: iter_time=1.144e-04, loss_ctc=74.968, loss_att=60.908, acc=0.682, loss=65.128, grad_norm=3.881, loss_scale=1.000, learning_rate=1.060e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 21:14:23,719 (deepspeed_trainer:228) INFO: 15epoch:train:3501-3600batch: iter_time=1.126e-04, loss_ctc=81.275, loss_att=59.004, acc=0.686, loss=65.707, grad_norm=4.136, loss_scale=1.000, learning_rate=1.060e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-05 21:14:59,988 (deepspeed_trainer:228) INFO: 15epoch:train:3601-3700batch: iter_time=1.120e-04, loss_ctc=84.649, loss_att=54.026, acc=0.692, loss=63.223, grad_norm=4.409, loss_scale=1.000, learning_rate=1.060e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 21:15:21,060 (multiple_iter_factory:32) INFO: Building 2th iter-factory... [cnode7-012:0/16] 2024-12-05 21:15:47,000 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-05 21:16:02,451 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.4", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.4", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.4", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.4", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-05 21:16:02,452 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.4, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-05 21:16:02,454 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-05 21:16:29,583] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:16:29,886] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:16:30,309] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:16:30,680] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:16:31,672] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:16:32,459] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:16:32,491] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:16:32,958] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:16:33,076] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:16:33,223] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:16:33,349] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:16:33,503] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:16:33,532] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:16:33,611] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:16:33,607] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:16:34,226] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:17:16,924] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:17:17,676] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:17:18,629] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:17:18,957] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:17:19,961] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:17:20,560] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:17:21,042] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:17:21,098] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:17:21,152] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:17:21,404] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:17:21,500] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:17:21,712] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:17:21,784] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:17:22,182] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:17:22,610] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:17:22,988] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:18:04,699] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:18:05,633] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:18:06,382] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:18:07,353] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:18:08,149] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:18:08,408] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:18:08,632] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:18:09,105] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:18:09,193] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:18:09,506] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:18:09,867] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:18:10,003] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:18:10,257] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:18:10,887] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:18:11,278] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:18:14,824] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:18:51,901] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:18:52,375] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:18:53,350] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:18:55,217] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:18:56,529] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:18:56,813] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:18:56,959] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:18:57,261] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:18:57,660] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:18:57,703] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:18:57,886] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:18:58,652] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:18:58,734] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:18:58,748] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:18:59,068] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:19:05,976] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-05 21:19:49,499 (deepspeed_trainer:228) INFO: 15epoch:train:3701-3800batch: iter_time=2.453, loss_ctc=88.427, loss_att=69.290, acc=0.673, loss=74.986, grad_norm=4.684, loss_scale=1.000, learning_rate=1.060e-04, step_time=0.442 [cnode7-012:0/16] 2024-12-05 21:20:25,676 (deepspeed_trainer:228) INFO: 15epoch:train:3801-3900batch: iter_time=1.094e-04, loss_ctc=76.392, loss_att=56.632, acc=0.678, loss=62.537, grad_norm=4.572, loss_scale=1.000, learning_rate=1.059e-04, step_time=0.361 [2024-12-05 21:21:02,209] [INFO] [logging.py:129:log_dist] [Rank 0] step=214000, skipped=0, lr=[np.float64(0.00010590042901211133)], mom=[[0.9, 0.98]] [2024-12-05 21:21:02,210] [INFO] [timer.py:264:stop] epoch=0/micro_step=34000/global_step=34000, RunningAvgSamplesPerSec=45.11907605884629, CurrSamplesPerSec=46.152043029535776, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 21:21:02,212 (deepspeed_trainer:228) INFO: 15epoch:train:3901-4000batch: iter_time=1.141e-04, loss_ctc=85.108, loss_att=73.182, acc=0.664, loss=76.770, grad_norm=4.341, loss_scale=1.000, learning_rate=1.059e-04, step_time=0.365 [cnode7-012:0/16] 2024-12-05 21:21:38,983 (deepspeed_trainer:228) INFO: 15epoch:train:4001-4100batch: iter_time=1.109e-04, loss_ctc=79.037, loss_att=65.473, acc=0.672, loss=69.511, grad_norm=4.306, loss_scale=1.000, learning_rate=1.059e-04, step_time=0.367 [cnode7-012:0/16] 2024-12-05 21:22:15,342 (deepspeed_trainer:228) INFO: 15epoch:train:4101-4200batch: iter_time=1.139e-04, loss_ctc=78.039, loss_att=62.801, acc=0.679, loss=67.379, grad_norm=4.410, loss_scale=1.000, learning_rate=1.059e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-05 21:22:51,353 (deepspeed_trainer:228) INFO: 15epoch:train:4201-4300batch: iter_time=1.128e-04, loss_ctc=86.567, loss_att=68.113, acc=0.683, loss=73.646, grad_norm=4.395, loss_scale=1.000, learning_rate=1.058e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 21:23:28,228 (deepspeed_trainer:228) INFO: 15epoch:train:4301-4400batch: iter_time=1.136e-04, loss_ctc=69.392, loss_att=56.429, acc=0.675, loss=60.334, grad_norm=4.270, loss_scale=1.000, learning_rate=1.058e-04, step_time=0.368 [cnode7-012:0/16] 2024-12-05 21:24:04,549 (deepspeed_trainer:228) INFO: 15epoch:train:4401-4500batch: iter_time=1.123e-04, loss_ctc=68.579, loss_att=54.435, acc=0.690, loss=58.701, grad_norm=3.726, loss_scale=1.000, learning_rate=1.058e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-05 21:24:40,963 (deepspeed_trainer:228) INFO: 15epoch:train:4501-4600batch: iter_time=1.113e-04, loss_ctc=74.990, loss_att=60.087, acc=0.693, loss=64.547, grad_norm=3.812, loss_scale=1.000, learning_rate=1.058e-04, step_time=0.364 [cnode7-012:0/16] 2024-12-05 21:25:17,175 (deepspeed_trainer:228) INFO: 15epoch:train:4601-4700batch: iter_time=1.133e-04, loss_ctc=78.725, loss_att=74.059, acc=0.669, loss=75.467, grad_norm=4.107, loss_scale=1.000, learning_rate=1.057e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 21:25:53,359 (deepspeed_trainer:228) INFO: 15epoch:train:4701-4800batch: iter_time=1.109e-04, loss_ctc=85.162, loss_att=66.869, acc=0.691, loss=72.344, grad_norm=4.368, loss_scale=1.000, learning_rate=1.057e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 21:26:29,308 (deepspeed_trainer:228) INFO: 15epoch:train:4801-4900batch: iter_time=1.122e-04, loss_ctc=78.089, loss_att=57.712, acc=0.690, loss=63.827, grad_norm=4.149, loss_scale=1.000, learning_rate=1.057e-04, step_time=0.359 [2024-12-05 21:27:05,585] [INFO] [logging.py:129:log_dist] [Rank 0] step=215000, skipped=0, lr=[np.float64(0.0001056538630527023)], mom=[[0.9, 0.98]] [2024-12-05 21:27:05,586] [INFO] [timer.py:264:stop] epoch=0/micro_step=35000/global_step=35000, RunningAvgSamplesPerSec=45.12237154916299, CurrSamplesPerSec=43.02713021244423, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 21:27:05,588 (deepspeed_trainer:228) INFO: 15epoch:train:4901-5000batch: iter_time=1.131e-04, loss_ctc=86.313, loss_att=65.538, acc=0.696, loss=71.771, grad_norm=4.700, loss_scale=1.000, learning_rate=1.057e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 21:27:42,407 (deepspeed_trainer:228) INFO: 15epoch:train:5001-5100batch: iter_time=1.112e-04, loss_ctc=83.037, loss_att=62.905, acc=0.701, loss=68.944, grad_norm=3.826, loss_scale=1.000, learning_rate=1.056e-04, step_time=0.368 [cnode7-012:0/16] 2024-12-05 21:28:18,691 (deepspeed_trainer:228) INFO: 15epoch:train:5101-5200batch: iter_time=1.156e-04, loss_ctc=81.358, loss_att=61.420, acc=0.689, loss=67.413, grad_norm=3.967, loss_scale=1.000, learning_rate=1.056e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-05 21:28:54,518 (deepspeed_trainer:228) INFO: 15epoch:train:5201-5300batch: iter_time=1.175e-04, loss_ctc=76.387, loss_att=64.079, acc=0.673, loss=67.768, grad_norm=4.242, loss_scale=1.000, learning_rate=1.056e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 21:29:30,328 (deepspeed_trainer:228) INFO: 15epoch:train:5301-5400batch: iter_time=1.134e-04, loss_ctc=79.239, loss_att=64.236, acc=0.685, loss=68.732, grad_norm=4.375, loss_scale=1.000, learning_rate=1.056e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 21:30:06,680 (deepspeed_trainer:228) INFO: 15epoch:train:5401-5500batch: iter_time=1.188e-04, loss_ctc=74.013, loss_att=56.237, acc=0.697, loss=61.578, grad_norm=3.821, loss_scale=1.000, learning_rate=1.055e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-05 21:30:42,309 (deepspeed_trainer:228) INFO: 15epoch:train:5501-5600batch: iter_time=1.127e-04, loss_ctc=92.219, loss_att=59.353, acc=0.680, loss=69.223, grad_norm=4.977, loss_scale=1.000, learning_rate=1.055e-04, step_time=0.356 [cnode7-012:0/16] 2024-12-05 21:30:54,128 (multiple_iter_factory:32) INFO: Building 3th iter-factory... [cnode7-012:0/16] 2024-12-05 21:31:20,165 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-05 21:31:36,272 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.2", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.2", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.2", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.2", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-05 21:31:36,272 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.2, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-05 21:31:36,274 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-05 21:32:02,685] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:32:03,725] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:32:03,770] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:32:03,870] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:32:04,507] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:32:04,647] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:32:05,271] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:32:05,510] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:32:05,846] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:32:05,866] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:32:06,387] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:32:06,455] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:32:06,518] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:32:06,631] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:32:06,798] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:32:07,041] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:32:50,720] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:32:51,649] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:32:52,400] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:32:52,363] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:32:52,748] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:32:53,328] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:32:53,832] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:32:54,288] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:32:54,584] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:32:55,101] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:32:55,288] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:32:55,454] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:32:55,593] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:32:56,041] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:32:58,135] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:32:58,204] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:33:37,985] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:33:39,097] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:33:39,904] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:33:40,710] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:33:41,086] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:33:41,392] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:33:42,368] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:33:43,086] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:33:43,137] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:33:43,162] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:33:43,426] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:33:43,600] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:33:44,098] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:33:44,265] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:33:48,559] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:33:49,401] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:34:25,677] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:34:27,253] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:34:27,645] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:34:28,820] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:34:29,048] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:34:29,729] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:34:29,924] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:34:30,944] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:34:31,413] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:34:31,657] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:34:31,846] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:34:32,447] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:34:32,496] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:34:35,697] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:34:37,939] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:34:38,789] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-05 21:35:32,175 (deepspeed_trainer:228) INFO: 15epoch:train:5601-5700batch: iter_time=2.418, loss_ctc=80.762, loss_att=65.323, acc=0.674, loss=69.937, grad_norm=4.483, loss_scale=1.000, learning_rate=1.055e-04, step_time=0.481 [cnode7-012:0/16] 2024-12-05 21:36:08,347 (deepspeed_trainer:228) INFO: 15epoch:train:5701-5800batch: iter_time=1.065e-04, loss_ctc=83.788, loss_att=58.262, acc=0.698, loss=65.911, grad_norm=4.284, loss_scale=1.000, learning_rate=1.055e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 21:36:44,771 (deepspeed_trainer:228) INFO: 15epoch:train:5801-5900batch: iter_time=1.100e-04, loss_ctc=75.252, loss_att=66.391, acc=0.668, loss=69.024, grad_norm=4.069, loss_scale=1.000, learning_rate=1.054e-04, step_time=0.364 [2024-12-05 21:37:21,190] [INFO] [logging.py:129:log_dist] [Rank 0] step=216000, skipped=0, lr=[np.float64(0.00010540901133688734)], mom=[[0.9, 0.98]] [2024-12-05 21:37:21,190] [INFO] [timer.py:264:stop] epoch=0/micro_step=36000/global_step=36000, RunningAvgSamplesPerSec=45.089583519133946, CurrSamplesPerSec=45.770444860452315, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 21:37:21,193 (deepspeed_trainer:228) INFO: 15epoch:train:5901-6000batch: iter_time=1.105e-04, loss_ctc=83.802, loss_att=68.282, acc=0.671, loss=72.954, grad_norm=4.310, loss_scale=1.000, learning_rate=1.054e-04, step_time=0.364 [cnode7-012:0/16] 2024-12-05 21:37:57,582 (deepspeed_trainer:228) INFO: 15epoch:train:6001-6100batch: iter_time=1.130e-04, loss_ctc=80.606, loss_att=63.681, acc=0.686, loss=68.765, grad_norm=4.211, loss_scale=1.000, learning_rate=1.054e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-05 21:38:34,117 (deepspeed_trainer:228) INFO: 15epoch:train:6101-6200batch: iter_time=1.106e-04, loss_ctc=77.023, loss_att=61.905, acc=0.687, loss=66.448, grad_norm=4.071, loss_scale=1.000, learning_rate=1.054e-04, step_time=0.365 [cnode7-012:0/16] 2024-12-05 21:39:10,280 (deepspeed_trainer:228) INFO: 15epoch:train:6201-6300batch: iter_time=1.134e-04, loss_ctc=70.174, loss_att=54.937, acc=0.687, loss=59.520, grad_norm=4.062, loss_scale=1.000, learning_rate=1.053e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 21:39:46,517 (deepspeed_trainer:228) INFO: 15epoch:train:6301-6400batch: iter_time=1.121e-04, loss_ctc=69.797, loss_att=55.673, acc=0.689, loss=59.935, grad_norm=3.800, loss_scale=1.000, learning_rate=1.053e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 21:40:22,785 (deepspeed_trainer:228) INFO: 15epoch:train:6401-6500batch: iter_time=1.113e-04, loss_ctc=78.338, loss_att=67.161, acc=0.680, loss=70.539, grad_norm=3.980, loss_scale=1.000, learning_rate=1.053e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 21:40:59,025 (deepspeed_trainer:228) INFO: 15epoch:train:6501-6600batch: iter_time=1.124e-04, loss_ctc=75.077, loss_att=67.708, acc=0.680, loss=69.934, grad_norm=3.910, loss_scale=1.000, learning_rate=1.053e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 21:41:35,182 (deepspeed_trainer:228) INFO: 15epoch:train:6601-6700batch: iter_time=1.120e-04, loss_ctc=83.357, loss_att=63.096, acc=0.695, loss=69.192, grad_norm=4.306, loss_scale=1.000, learning_rate=1.053e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 21:42:11,647 (deepspeed_trainer:228) INFO: 15epoch:train:6701-6800batch: iter_time=1.132e-04, loss_ctc=80.891, loss_att=57.970, acc=0.699, loss=64.886, grad_norm=4.236, loss_scale=1.000, learning_rate=1.052e-04, step_time=0.364 [cnode7-012:0/16] 2024-12-05 21:42:47,901 (deepspeed_trainer:228) INFO: 15epoch:train:6801-6900batch: iter_time=1.108e-04, loss_ctc=89.881, loss_att=69.517, acc=0.691, loss=75.626, grad_norm=4.488, loss_scale=1.000, learning_rate=1.052e-04, step_time=0.362 [2024-12-05 21:43:24,175] [INFO] [logging.py:129:log_dist] [Rank 0] step=217000, skipped=0, lr=[np.float64(0.0001051658540925668)], mom=[[0.9, 0.98]] [2024-12-05 21:43:24,175] [INFO] [timer.py:264:stop] epoch=0/micro_step=37000/global_step=37000, RunningAvgSamplesPerSec=45.095823525161066, CurrSamplesPerSec=43.623578535978005, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 21:43:24,177 (deepspeed_trainer:228) INFO: 15epoch:train:6901-7000batch: iter_time=1.108e-04, loss_ctc=77.688, loss_att=58.233, acc=0.704, loss=64.082, grad_norm=3.747, loss_scale=1.000, learning_rate=1.052e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 21:44:00,296 (deepspeed_trainer:228) INFO: 15epoch:train:7001-7100batch: iter_time=1.124e-04, loss_ctc=78.388, loss_att=61.047, acc=0.681, loss=66.255, grad_norm=4.043, loss_scale=1.000, learning_rate=1.052e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 21:44:36,368 (deepspeed_trainer:228) INFO: 15epoch:train:7101-7200batch: iter_time=1.106e-04, loss_ctc=71.973, loss_att=59.376, acc=0.687, loss=63.155, grad_norm=3.799, loss_scale=1.000, learning_rate=1.051e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 21:45:12,819 (deepspeed_trainer:228) INFO: 15epoch:train:7201-7300batch: iter_time=1.095e-04, loss_ctc=85.846, loss_att=69.578, acc=0.677, loss=74.481, grad_norm=4.362, loss_scale=1.000, learning_rate=1.051e-04, step_time=0.364 [cnode7-012:0/16] 2024-12-05 21:45:48,950 (deepspeed_trainer:228) INFO: 15epoch:train:7301-7400batch: iter_time=1.110e-04, loss_ctc=67.946, loss_att=52.519, acc=0.698, loss=57.127, grad_norm=3.738, loss_scale=1.000, learning_rate=1.051e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 21:46:26,315 (deepspeed_trainer:228) INFO: 15epoch:train:7401-7500batch: iter_time=1.084e-04, loss_ctc=102.023, loss_att=66.386, acc=0.689, loss=77.078, grad_norm=5.035, loss_scale=1.000, learning_rate=1.051e-04, step_time=0.373 [cnode7-012:0/16] 2024-12-05 21:46:28,877 (multiple_iter_factory:32) INFO: Building 4th iter-factory... [cnode7-012:0/16] 2024-12-05 21:46:55,437 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-05 21:47:10,963 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.0", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.0", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.0", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.0", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-05 21:47:10,963 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.0, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-05 21:47:10,965 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-05 21:47:38,765] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:47:38,801] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:47:39,069] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:47:39,616] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:47:39,778] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:47:40,570] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:47:40,636] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:47:40,824] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:47:40,881] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:47:41,358] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:47:43,327] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:47:43,287] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:47:43,387] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:47:44,075] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:47:44,137] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:47:44,174] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:48:27,225] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:48:27,412] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:48:27,689] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:48:29,535] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:48:29,863] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:48:29,888] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:48:29,956] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:48:30,713] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:48:30,664] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:48:30,730] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:48:30,777] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:48:30,932] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:48:31,537] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:48:32,099] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:48:32,184] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:48:32,592] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:49:15,134] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:49:15,334] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:49:17,470] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:49:17,839] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:49:17,770] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:49:18,303] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:49:18,552] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:49:18,713] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:49:18,985] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:49:19,100] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:49:19,362] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:49:19,590] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:49:19,643] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:49:20,502] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:49:20,610] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:49:24,725] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:50:02,963] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:50:03,007] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:50:05,269] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:50:05,526] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:50:05,988] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:50:06,347] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:50:06,543] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:50:07,118] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:50:07,479] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:50:07,733] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:50:08,473] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:50:08,600] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:50:08,746] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:50:09,348] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:50:09,776] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 21:50:14,288] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-05 21:51:16,612 (deepspeed_trainer:228) INFO: 15epoch:train:7501-7600batch: iter_time=2.439, loss_ctc=72.144, loss_att=58.254, acc=0.673, loss=62.430, grad_norm=4.410, loss_scale=1.000, learning_rate=1.050e-04, step_time=0.464 [cnode7-012:0/16] 2024-12-05 21:51:52,614 (deepspeed_trainer:228) INFO: 15epoch:train:7601-7700batch: iter_time=1.083e-04, loss_ctc=81.612, loss_att=60.363, acc=0.692, loss=66.757, grad_norm=3.843, loss_scale=1.000, learning_rate=1.050e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 21:52:29,216 (deepspeed_trainer:228) INFO: 15epoch:train:7701-7800batch: iter_time=1.133e-04, loss_ctc=80.258, loss_att=71.426, acc=0.664, loss=74.081, grad_norm=4.092, loss_scale=1.000, learning_rate=1.050e-04, step_time=0.366 [cnode7-012:0/16] 2024-12-05 21:53:05,251 (deepspeed_trainer:228) INFO: 15epoch:train:7801-7900batch: iter_time=1.148e-04, loss_ctc=78.360, loss_att=62.279, acc=0.682, loss=67.091, grad_norm=4.088, loss_scale=1.000, learning_rate=1.050e-04, step_time=0.360 [2024-12-05 21:53:41,047] [INFO] [logging.py:129:log_dist] [Rank 0] step=218000, skipped=0, lr=[np.float64(0.00010492437186544728)], mom=[[0.9, 0.98]] [2024-12-05 21:53:41,047] [INFO] [timer.py:264:stop] epoch=0/micro_step=38000/global_step=38000, RunningAvgSamplesPerSec=45.06845098776406, CurrSamplesPerSec=46.22414098394458, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 21:53:41,049 (deepspeed_trainer:228) INFO: 15epoch:train:7901-8000batch: iter_time=1.115e-04, loss_ctc=77.494, loss_att=61.411, acc=0.688, loss=66.233, grad_norm=4.162, loss_scale=1.000, learning_rate=1.049e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 21:54:17,340 (deepspeed_trainer:228) INFO: 15epoch:train:8001-8100batch: iter_time=1.124e-04, loss_ctc=75.414, loss_att=60.448, acc=0.685, loss=64.955, grad_norm=4.316, loss_scale=1.000, learning_rate=1.049e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 21:54:53,612 (deepspeed_trainer:228) INFO: 15epoch:train:8101-8200batch: iter_time=1.137e-04, loss_ctc=68.681, loss_att=54.243, acc=0.683, loss=58.555, grad_norm=3.863, loss_scale=1.000, learning_rate=1.049e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 21:55:29,499 (deepspeed_trainer:228) INFO: 15epoch:train:8201-8300batch: iter_time=1.120e-04, loss_ctc=72.547, loss_att=57.886, acc=0.692, loss=62.268, grad_norm=3.728, loss_scale=1.000, learning_rate=1.049e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 21:56:05,663 (deepspeed_trainer:228) INFO: 15epoch:train:8301-8400batch: iter_time=1.132e-04, loss_ctc=81.716, loss_att=71.491, acc=0.682, loss=74.561, grad_norm=3.999, loss_scale=1.000, learning_rate=1.048e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 21:56:41,531 (deepspeed_trainer:228) INFO: 15epoch:train:8401-8500batch: iter_time=1.101e-04, loss_ctc=74.706, loss_att=65.642, acc=0.687, loss=68.355, grad_norm=3.970, loss_scale=1.000, learning_rate=1.048e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 21:57:17,726 (deepspeed_trainer:228) INFO: 15epoch:train:8501-8600batch: iter_time=1.136e-04, loss_ctc=81.018, loss_att=57.113, acc=0.698, loss=64.289, grad_norm=4.169, loss_scale=1.000, learning_rate=1.048e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 21:57:54,011 (deepspeed_trainer:228) INFO: 15epoch:train:8601-8700batch: iter_time=1.140e-04, loss_ctc=84.263, loss_att=64.760, acc=0.691, loss=70.575, grad_norm=4.691, loss_scale=1.000, learning_rate=1.048e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 21:58:30,156 (deepspeed_trainer:228) INFO: 15epoch:train:8701-8800batch: iter_time=1.115e-04, loss_ctc=83.281, loss_att=61.454, acc=0.710, loss=68.032, grad_norm=3.832, loss_scale=1.000, learning_rate=1.047e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 21:59:06,180 (deepspeed_trainer:228) INFO: 15epoch:train:8801-8900batch: iter_time=1.130e-04, loss_ctc=79.034, loss_att=58.877, acc=0.699, loss=64.914, grad_norm=3.634, loss_scale=1.000, learning_rate=1.047e-04, step_time=0.360 [2024-12-05 21:59:42,250] [INFO] [logging.py:129:log_dist] [Rank 0] step=219000, skipped=0, lr=[np.float64(0.00010468454551250392)], mom=[[0.9, 0.98]] [2024-12-05 21:59:42,250] [INFO] [timer.py:264:stop] epoch=0/micro_step=39000/global_step=39000, RunningAvgSamplesPerSec=45.08108924332829, CurrSamplesPerSec=45.34976556951327, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 21:59:42,252 (deepspeed_trainer:228) INFO: 15epoch:train:8901-9000batch: iter_time=1.145e-04, loss_ctc=80.027, loss_att=63.051, acc=0.678, loss=68.137, grad_norm=4.100, loss_scale=1.000, learning_rate=1.047e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 22:00:18,115 (deepspeed_trainer:228) INFO: 15epoch:train:9001-9100batch: iter_time=1.104e-04, loss_ctc=71.656, loss_att=59.421, acc=0.695, loss=63.115, grad_norm=3.793, loss_scale=1.000, learning_rate=1.047e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 22:00:54,118 (deepspeed_trainer:228) INFO: 15epoch:train:9101-9200batch: iter_time=1.138e-04, loss_ctc=82.383, loss_att=66.892, acc=0.684, loss=71.507, grad_norm=4.245, loss_scale=1.000, learning_rate=1.046e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 22:01:30,086 (deepspeed_trainer:228) INFO: 15epoch:train:9201-9300batch: iter_time=1.125e-04, loss_ctc=81.571, loss_att=52.422, acc=0.698, loss=61.163, grad_norm=4.241, loss_scale=1.000, learning_rate=1.046e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 22:01:59,755 (multiple_iter_factory:32) INFO: Building 5th iter-factory... [cnode7-012:0/16] 2024-12-05 22:02:26,643 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-05 22:02:43,190 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.5", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.5", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.5", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.5", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-05 22:02:43,190 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.5, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-05 22:02:43,193 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-05 22:03:09,519] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:03:10,153] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:03:10,653] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:03:10,898] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:03:11,104] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:03:11,325] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:03:11,528] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:03:11,999] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:03:12,051] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:03:12,607] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:03:12,518] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:03:12,586] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:03:12,727] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:03:12,633] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:03:13,041] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:03:13,441] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:03:57,462] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:03:58,774] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:03:59,095] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:03:59,366] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:03:59,322] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:03:59,683] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:04:00,018] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:04:00,327] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:04:00,448] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:04:00,950] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:04:01,314] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:04:01,361] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:04:01,423] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:04:01,724] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:04:02,220] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:04:02,262] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:04:46,333] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:04:47,003] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:04:47,236] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:04:47,915] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:04:48,137] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:04:48,942] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:04:49,704] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:04:49,909] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:04:50,385] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:04:50,315] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:04:50,353] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:04:50,531] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:04:50,995] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:04:51,108] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:04:52,973] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:04:53,580] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:05:33,913] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:05:34,301] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:05:37,600] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:05:38,032] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:05:37,986] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:05:38,374] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:05:38,388] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:05:38,549] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:05:39,339] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:05:39,408] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:05:39,772] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:05:39,712] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:05:40,032] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:05:40,409] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:05:43,768] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:05:43,771] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-05 22:06:19,266 (deepspeed_trainer:228) INFO: 15epoch:train:9301-9400batch: iter_time=2.488, loss_ctc=86.350, loss_att=64.078, acc=0.686, loss=70.747, grad_norm=4.582, loss_scale=1.000, learning_rate=1.046e-04, step_time=0.404 [cnode7-012:0/16] 2024-12-05 22:06:55,287 (deepspeed_trainer:228) INFO: 15epoch:train:9401-9500batch: iter_time=1.113e-04, loss_ctc=80.326, loss_att=61.898, acc=0.667, loss=67.396, grad_norm=4.770, loss_scale=1.000, learning_rate=1.046e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 22:07:31,381 (deepspeed_trainer:228) INFO: 15epoch:train:9501-9600batch: iter_time=1.124e-04, loss_ctc=79.847, loss_att=64.448, acc=0.683, loss=69.051, grad_norm=4.485, loss_scale=1.000, learning_rate=1.046e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 22:08:07,227 (deepspeed_trainer:228) INFO: 15epoch:train:9601-9700batch: iter_time=1.130e-04, loss_ctc=75.914, loss_att=64.416, acc=0.665, loss=67.862, grad_norm=4.496, loss_scale=1.000, learning_rate=1.045e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 22:08:42,965 (deepspeed_trainer:228) INFO: 15epoch:train:9701-9800batch: iter_time=1.145e-04, loss_ctc=79.009, loss_att=64.478, acc=0.678, loss=68.861, grad_norm=3.910, loss_scale=1.000, learning_rate=1.045e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-05 22:09:18,847 (deepspeed_trainer:228) INFO: 15epoch:train:9801-9900batch: iter_time=1.130e-04, loss_ctc=84.814, loss_att=66.769, acc=0.676, loss=72.226, grad_norm=4.381, loss_scale=1.000, learning_rate=1.045e-04, step_time=0.358 [2024-12-05 22:09:54,771] [INFO] [logging.py:129:log_dist] [Rank 0] step=220000, skipped=0, lr=[np.float64(0.0001044463561956062)], mom=[[0.9, 0.98]] [2024-12-05 22:09:54,774] [INFO] [timer.py:264:stop] epoch=0/micro_step=40000/global_step=40000, RunningAvgSamplesPerSec=45.08432287719205, CurrSamplesPerSec=44.897034963101326, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 22:09:54,776 (deepspeed_trainer:228) INFO: 15epoch:train:9901-10000batch: iter_time=1.098e-04, loss_ctc=67.600, loss_att=52.377, acc=0.681, loss=56.958, grad_norm=4.202, loss_scale=1.000, learning_rate=1.045e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 22:10:31,301 (deepspeed_trainer:228) INFO: 15epoch:train:10001-10100batch: iter_time=1.122e-04, loss_ctc=70.426, loss_att=61.006, acc=0.679, loss=63.842, grad_norm=3.928, loss_scale=1.000, learning_rate=1.044e-04, step_time=0.365 [cnode7-012:0/16] 2024-12-05 22:11:07,317 (deepspeed_trainer:228) INFO: 15epoch:train:10101-10200batch: iter_time=1.130e-04, loss_ctc=73.675, loss_att=57.628, acc=0.690, loss=62.429, grad_norm=3.719, loss_scale=1.000, learning_rate=1.044e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 22:11:43,876 (deepspeed_trainer:228) INFO: 15epoch:train:10201-10300batch: iter_time=1.105e-04, loss_ctc=76.613, loss_att=71.009, acc=0.672, loss=72.692, grad_norm=3.915, loss_scale=1.000, learning_rate=1.044e-04, step_time=0.365 [cnode7-012:0/16] 2024-12-05 22:12:20,430 (deepspeed_trainer:228) INFO: 15epoch:train:10301-10400batch: iter_time=1.095e-04, loss_ctc=80.131, loss_att=64.957, acc=0.686, loss=69.535, grad_norm=4.345, loss_scale=1.000, learning_rate=1.044e-04, step_time=0.365 [cnode7-012:0/16] 2024-12-05 22:12:56,496 (deepspeed_trainer:228) INFO: 15epoch:train:10401-10500batch: iter_time=1.096e-04, loss_ctc=80.321, loss_att=56.569, acc=0.690, loss=63.677, grad_norm=4.148, loss_scale=1.000, learning_rate=1.043e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 22:13:32,568 (deepspeed_trainer:228) INFO: 15epoch:train:10501-10600batch: iter_time=1.110e-04, loss_ctc=82.113, loss_att=61.209, acc=0.697, loss=67.473, grad_norm=4.254, loss_scale=1.000, learning_rate=1.043e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 22:14:08,852 (deepspeed_trainer:228) INFO: 15epoch:train:10601-10700batch: iter_time=1.130e-04, loss_ctc=86.751, loss_att=66.054, acc=0.698, loss=72.260, grad_norm=4.143, loss_scale=1.000, learning_rate=1.043e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 22:14:44,733 (deepspeed_trainer:228) INFO: 15epoch:train:10701-10800batch: iter_time=1.125e-04, loss_ctc=74.046, loss_att=54.554, acc=0.688, loss=60.396, grad_norm=4.080, loss_scale=1.000, learning_rate=1.043e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 22:15:20,592 (deepspeed_trainer:228) INFO: 15epoch:train:10801-10900batch: iter_time=1.111e-04, loss_ctc=80.211, loss_att=66.778, acc=0.664, loss=70.808, grad_norm=4.143, loss_scale=1.000, learning_rate=1.042e-04, step_time=0.358 [2024-12-05 22:15:56,394] [INFO] [logging.py:129:log_dist] [Rank 0] step=221000, skipped=0, lr=[np.float64(0.00010420978537530275)], mom=[[0.9, 0.98]] [2024-12-05 22:15:56,395] [INFO] [timer.py:264:stop] epoch=0/micro_step=41000/global_step=41000, RunningAvgSamplesPerSec=45.09417758259485, CurrSamplesPerSec=47.30973197569957, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 22:15:56,397 (deepspeed_trainer:228) INFO: 15epoch:train:10901-11000batch: iter_time=1.147e-04, loss_ctc=73.955, loss_att=61.399, acc=0.682, loss=65.204, grad_norm=4.058, loss_scale=1.000, learning_rate=1.042e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 22:16:32,515 (deepspeed_trainer:228) INFO: 15epoch:train:11001-11100batch: iter_time=1.127e-04, loss_ctc=79.377, loss_att=58.097, acc=0.693, loss=64.476, grad_norm=3.911, loss_scale=1.000, learning_rate=1.042e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 22:17:08,691 (deepspeed_trainer:228) INFO: 15epoch:train:11101-11200batch: iter_time=1.110e-04, loss_ctc=82.858, loss_att=52.759, acc=0.697, loss=61.803, grad_norm=5.059, loss_scale=1.000, learning_rate=1.042e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 22:17:30,938 (multiple_iter_factory:32) INFO: Building 6th iter-factory... [cnode7-012:0/16] 2024-12-05 22:17:57,353 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-05 22:18:12,591 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.6", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.6", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.6", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.6", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-05 22:18:12,592 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.6, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-05 22:18:12,594 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-05 22:18:38,387] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:18:39,426] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:18:40,380] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:18:40,255] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:18:40,632] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:18:40,832] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:18:40,963] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:18:41,411] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:18:41,292] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:18:41,939] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:18:41,957] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:18:42,015] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:18:42,530] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:18:42,772] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:18:42,918] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:18:44,833] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:19:25,697] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:19:26,812] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:19:27,078] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:19:28,993] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:19:29,339] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:19:29,487] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:19:29,558] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:19:29,633] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:19:29,986] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:19:30,226] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:19:30,699] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:19:30,859] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:19:30,869] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:19:31,144] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:19:31,409] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:19:33,673] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:20:12,233] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:20:14,214] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:20:14,790] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:20:16,921] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:20:17,232] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:20:17,450] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:20:17,717] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:20:18,372] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:20:18,711] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:20:18,643] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:20:18,868] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:20:19,041] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:20:19,141] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:20:19,590] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:20:19,638] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:20:21,026] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:20:58,520] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:21:01,820] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:21:01,950] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:21:05,606] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:21:05,521] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:21:06,086] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:21:05,985] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:21:06,078] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:21:06,477] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:21:06,715] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:21:06,667] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:21:07,001] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:21:07,544] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:21:07,632] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:21:07,817] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:21:09,681] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-05 22:21:54,395 (deepspeed_trainer:228) INFO: 15epoch:train:11201-11300batch: iter_time=2.407, loss_ctc=86.131, loss_att=67.174, acc=0.677, loss=72.847, grad_norm=4.726, loss_scale=1.000, learning_rate=1.042e-04, step_time=0.450 [cnode7-012:0/16] 2024-12-05 22:22:30,360 (deepspeed_trainer:228) INFO: 15epoch:train:11301-11400batch: iter_time=1.073e-04, loss_ctc=74.326, loss_att=54.380, acc=0.684, loss=60.358, grad_norm=4.600, loss_scale=1.000, learning_rate=1.041e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 22:23:06,567 (deepspeed_trainer:228) INFO: 15epoch:train:11401-11500batch: iter_time=1.075e-04, loss_ctc=82.567, loss_att=71.249, acc=0.672, loss=74.633, grad_norm=4.271, loss_scale=1.000, learning_rate=1.041e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 22:23:43,711 (deepspeed_trainer:228) INFO: 15epoch:train:11501-11600batch: iter_time=1.097e-04, loss_ctc=77.338, loss_att=63.732, acc=0.679, loss=67.829, grad_norm=4.165, loss_scale=1.000, learning_rate=1.041e-04, step_time=0.371 [cnode7-012:0/16] 2024-12-05 22:24:20,152 (deepspeed_trainer:228) INFO: 15epoch:train:11601-11700batch: iter_time=1.112e-04, loss_ctc=76.402, loss_att=61.758, acc=0.684, loss=66.151, grad_norm=4.319, loss_scale=1.000, learning_rate=1.041e-04, step_time=0.364 [cnode7-012:0/16] 2024-12-05 22:24:56,345 (deepspeed_trainer:228) INFO: 15epoch:train:11701-11800batch: iter_time=1.112e-04, loss_ctc=84.899, loss_att=66.803, acc=0.689, loss=72.218, grad_norm=4.245, loss_scale=1.000, learning_rate=1.040e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 22:25:32,220 (deepspeed_trainer:228) INFO: 15epoch:train:11801-11900batch: iter_time=1.085e-04, loss_ctc=68.384, loss_att=55.113, acc=0.682, loss=59.064, grad_norm=4.124, loss_scale=1.000, learning_rate=1.040e-04, step_time=0.358 [2024-12-05 22:26:08,030] [INFO] [logging.py:129:log_dist] [Rank 0] step=222000, skipped=0, lr=[np.float64(0.00010397481480476015)], mom=[[0.9, 0.98]] [2024-12-05 22:26:08,031] [INFO] [timer.py:264:stop] epoch=0/micro_step=42000/global_step=42000, RunningAvgSamplesPerSec=45.07468453801269, CurrSamplesPerSec=44.85313391462498, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 22:26:08,032 (deepspeed_trainer:228) INFO: 15epoch:train:11901-12000batch: iter_time=1.100e-04, loss_ctc=68.188, loss_att=53.484, acc=0.695, loss=57.921, grad_norm=3.592, loss_scale=1.000, learning_rate=1.040e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 22:26:44,114 (deepspeed_trainer:228) INFO: 15epoch:train:12001-12100batch: iter_time=1.167e-04, loss_ctc=74.143, loss_att=59.342, acc=0.697, loss=63.755, grad_norm=3.580, loss_scale=1.000, learning_rate=1.040e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 22:27:20,520 (deepspeed_trainer:228) INFO: 15epoch:train:12101-12200batch: iter_time=1.155e-04, loss_ctc=76.971, loss_att=71.789, acc=0.678, loss=73.344, grad_norm=4.107, loss_scale=1.000, learning_rate=1.039e-04, step_time=0.364 [cnode7-012:0/16] 2024-12-05 22:27:57,249 (deepspeed_trainer:228) INFO: 15epoch:train:12201-12300batch: iter_time=1.115e-04, loss_ctc=83.889, loss_att=65.614, acc=0.697, loss=71.075, grad_norm=4.081, loss_scale=1.000, learning_rate=1.039e-04, step_time=0.367 [cnode7-012:0/16] 2024-12-05 22:28:33,572 (deepspeed_trainer:228) INFO: 15epoch:train:12301-12400batch: iter_time=1.111e-04, loss_ctc=77.031, loss_att=56.699, acc=0.694, loss=62.800, grad_norm=3.951, loss_scale=1.000, learning_rate=1.039e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-05 22:29:10,184 (deepspeed_trainer:228) INFO: 15epoch:train:12401-12500batch: iter_time=1.112e-04, loss_ctc=86.612, loss_att=65.407, acc=0.699, loss=71.780, grad_norm=4.583, loss_scale=1.000, learning_rate=1.039e-04, step_time=0.366 [cnode7-012:0/16] 2024-12-05 22:29:46,706 (deepspeed_trainer:228) INFO: 15epoch:train:12501-12600batch: iter_time=1.102e-04, loss_ctc=81.864, loss_att=61.324, acc=0.708, loss=67.486, grad_norm=3.678, loss_scale=1.000, learning_rate=1.038e-04, step_time=0.365 [cnode7-012:0/16] 2024-12-05 22:30:22,620 (deepspeed_trainer:228) INFO: 15epoch:train:12601-12700batch: iter_time=1.118e-04, loss_ctc=80.901, loss_att=61.002, acc=0.691, loss=66.949, grad_norm=4.045, loss_scale=1.000, learning_rate=1.038e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 22:30:58,729 (deepspeed_trainer:228) INFO: 15epoch:train:12701-12800batch: iter_time=1.092e-04, loss_ctc=76.141, loss_att=64.089, acc=0.675, loss=67.748, grad_norm=3.926, loss_scale=1.000, learning_rate=1.038e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 22:31:34,817 (deepspeed_trainer:228) INFO: 15epoch:train:12801-12900batch: iter_time=1.107e-04, loss_ctc=77.750, loss_att=63.672, acc=0.688, loss=67.887, grad_norm=4.146, loss_scale=1.000, learning_rate=1.038e-04, step_time=0.361 [2024-12-05 22:32:10,517] [INFO] [logging.py:129:log_dist] [Rank 0] step=223000, skipped=0, lr=[np.float64(0.00010374142652385154)], mom=[[0.9, 0.98]] [2024-12-05 22:32:10,518] [INFO] [timer.py:264:stop] epoch=0/micro_step=43000/global_step=43000, RunningAvgSamplesPerSec=45.08117051129136, CurrSamplesPerSec=47.80330056517807, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 22:32:10,520 (deepspeed_trainer:228) INFO: 15epoch:train:12901-13000batch: iter_time=1.097e-04, loss_ctc=72.950, loss_att=54.862, acc=0.702, loss=60.279, grad_norm=3.804, loss_scale=1.000, learning_rate=1.038e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-05 22:32:46,097 (deepspeed_trainer:228) INFO: 15epoch:train:13001-13100batch: iter_time=1.119e-04, loss_ctc=89.999, loss_att=58.617, acc=0.686, loss=68.076, grad_norm=4.979, loss_scale=1.000, learning_rate=1.037e-04, step_time=0.355 [cnode7-012:0/16] 2024-12-05 22:32:57,766 (multiple_iter_factory:32) INFO: Building 7th iter-factory... [cnode7-012:0/16] 2024-12-05 22:33:24,733 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-05 22:33:40,799 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.7", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.7", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.7", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.7", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-05 22:33:40,799 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.7, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-05 22:33:40,801 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-05 22:34:08,402] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:34:09,114] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:34:09,307] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:34:09,592] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:34:09,575] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:34:10,013] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:34:10,063] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:34:09,991] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:34:10,468] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:34:10,512] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:34:10,577] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:34:11,441] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:34:11,520] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:34:11,747] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:34:11,972] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:34:12,126] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:34:57,030] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:34:57,795] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:34:58,137] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:34:58,119] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:34:58,624] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:34:58,728] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:34:58,944] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:34:59,155] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:34:59,203] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:34:59,833] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:34:59,789] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:34:59,944] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:35:00,365] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:35:00,586] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:35:01,026] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:35:03,606] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:35:45,633] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:35:45,815] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:35:46,269] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:35:47,048] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:35:47,420] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:35:47,612] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:35:47,706] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:35:48,077] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:35:48,430] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:35:48,533] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:35:48,617] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:35:48,819] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:35:49,138] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:35:49,850] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:35:52,324] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:35:54,203] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:36:34,439] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:36:34,486] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:36:34,885] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:36:35,617] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:36:36,043] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:36:36,288] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:36:36,653] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:36:36,802] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:36:36,926] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:36:37,370] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:36:37,414] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:36:37,468] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:36:38,773] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:36:41,579] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:36:43,703] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:36:44,442] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-05 22:37:37,274 (deepspeed_trainer:228) INFO: 15epoch:train:13101-13200batch: iter_time=2.455, loss_ctc=79.582, loss_att=64.011, acc=0.674, loss=68.699, grad_norm=4.573, loss_scale=1.000, learning_rate=1.037e-04, step_time=0.457 [cnode7-012:0/16] 2024-12-05 22:38:13,426 (deepspeed_trainer:228) INFO: 15epoch:train:13201-13300batch: iter_time=1.117e-04, loss_ctc=81.082, loss_att=57.281, acc=0.698, loss=64.399, grad_norm=4.370, loss_scale=1.000, learning_rate=1.037e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 22:38:49,013 (deepspeed_trainer:228) INFO: 15epoch:train:13301-13400batch: iter_time=1.114e-04, loss_ctc=73.508, loss_att=64.288, acc=0.672, loss=67.029, grad_norm=4.355, loss_scale=1.000, learning_rate=1.037e-04, step_time=0.356 [cnode7-012:0/16] 2024-12-05 22:39:24,768 (deepspeed_trainer:228) INFO: 15epoch:train:13401-13500batch: iter_time=1.095e-04, loss_ctc=81.118, loss_att=66.206, acc=0.671, loss=70.706, grad_norm=4.490, loss_scale=1.000, learning_rate=1.036e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-05 22:40:00,713 (deepspeed_trainer:228) INFO: 15epoch:train:13501-13600batch: iter_time=1.110e-04, loss_ctc=79.236, loss_att=62.903, acc=0.684, loss=67.818, grad_norm=4.098, loss_scale=1.000, learning_rate=1.036e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 22:40:36,302 (deepspeed_trainer:228) INFO: 15epoch:train:13601-13700batch: iter_time=1.115e-04, loss_ctc=76.939, loss_att=59.824, acc=0.688, loss=64.966, grad_norm=4.192, loss_scale=1.000, learning_rate=1.036e-04, step_time=0.356 [cnode7-012:0/16] 2024-12-05 22:41:12,080 (deepspeed_trainer:228) INFO: 15epoch:train:13701-13800batch: iter_time=1.131e-04, loss_ctc=69.154, loss_att=54.962, acc=0.683, loss=59.240, grad_norm=4.057, loss_scale=1.000, learning_rate=1.036e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-05 22:41:47,948 (deepspeed_trainer:228) INFO: 15epoch:train:13801-13900batch: iter_time=1.118e-04, loss_ctc=69.933, loss_att=55.295, acc=0.687, loss=59.692, grad_norm=4.230, loss_scale=1.000, learning_rate=1.035e-04, step_time=0.358 [2024-12-05 22:42:24,114] [INFO] [logging.py:129:log_dist] [Rank 0] step=224000, skipped=0, lr=[np.float64(0.00010350960285339035)], mom=[[0.9, 0.98]] [2024-12-05 22:42:24,115] [INFO] [timer.py:264:stop] epoch=0/micro_step=44000/global_step=44000, RunningAvgSamplesPerSec=45.0717973726874, CurrSamplesPerSec=47.524286426589306, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 22:42:24,117 (deepspeed_trainer:228) INFO: 15epoch:train:13901-14000batch: iter_time=1.118e-04, loss_ctc=77.841, loss_att=66.215, acc=0.673, loss=69.733, grad_norm=4.432, loss_scale=1.000, learning_rate=1.035e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 22:43:00,104 (deepspeed_trainer:228) INFO: 15epoch:train:14001-14100batch: iter_time=1.103e-04, loss_ctc=73.812, loss_att=66.839, acc=0.677, loss=68.961, grad_norm=4.203, loss_scale=1.000, learning_rate=1.035e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 22:43:36,176 (deepspeed_trainer:228) INFO: 15epoch:train:14101-14200batch: iter_time=1.122e-04, loss_ctc=82.390, loss_att=60.467, acc=0.696, loss=67.025, grad_norm=4.365, loss_scale=1.000, learning_rate=1.035e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 22:44:12,079 (deepspeed_trainer:228) INFO: 15epoch:train:14201-14300batch: iter_time=1.113e-04, loss_ctc=80.013, loss_att=57.157, acc=0.699, loss=64.017, grad_norm=4.276, loss_scale=1.000, learning_rate=1.035e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 22:44:48,304 (deepspeed_trainer:228) INFO: 15epoch:train:14301-14400batch: iter_time=1.108e-04, loss_ctc=90.638, loss_att=68.987, acc=0.688, loss=75.474, grad_norm=4.936, loss_scale=1.000, learning_rate=1.034e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 22:45:24,364 (deepspeed_trainer:228) INFO: 15epoch:train:14401-14500batch: iter_time=1.123e-04, loss_ctc=77.056, loss_att=56.827, acc=0.703, loss=62.911, grad_norm=4.028, loss_scale=1.000, learning_rate=1.034e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 22:46:00,294 (deepspeed_trainer:228) INFO: 15epoch:train:14501-14600batch: iter_time=1.120e-04, loss_ctc=77.827, loss_att=59.573, acc=0.678, loss=65.032, grad_norm=4.254, loss_scale=1.000, learning_rate=1.034e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 22:46:36,222 (deepspeed_trainer:228) INFO: 15epoch:train:14601-14700batch: iter_time=1.095e-04, loss_ctc=71.676, loss_att=58.347, acc=0.684, loss=62.340, grad_norm=4.309, loss_scale=1.000, learning_rate=1.034e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 22:47:12,323 (deepspeed_trainer:228) INFO: 15epoch:train:14701-14800batch: iter_time=1.107e-04, loss_ctc=84.910, loss_att=66.251, acc=0.678, loss=71.836, grad_norm=4.191, loss_scale=1.000, learning_rate=1.033e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 22:47:48,009 (deepspeed_trainer:228) INFO: 15epoch:train:14801-14900batch: iter_time=1.120e-04, loss_ctc=66.961, loss_att=51.040, acc=0.697, loss=55.808, grad_norm=4.229, loss_scale=1.000, learning_rate=1.033e-04, step_time=0.356 [2024-12-05 22:48:24,141] [INFO] [logging.py:129:log_dist] [Rank 0] step=225000, skipped=0, lr=[np.float64(0.00010327932638950527)], mom=[[0.9, 0.98]] [2024-12-05 22:48:24,142] [INFO] [timer.py:264:stop] epoch=0/micro_step=45000/global_step=45000, RunningAvgSamplesPerSec=45.085746824985264, CurrSamplesPerSec=46.245068600019295, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 22:48:24,143 (deepspeed_trainer:228) INFO: 15epoch:train:14901-15000batch: iter_time=1.084e-04, loss_ctc=100.579, loss_att=65.097, acc=0.690, loss=75.711, grad_norm=5.301, loss_scale=1.000, learning_rate=1.033e-04, step_time=0.361 [2024-12-05 22:48:38,160] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:48:39,306] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:48:39,445] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:48:39,563] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:48:39,668] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:48:39,803] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:48:39,964] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:48:40,079] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:48:40,134] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:48:40,156] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:48:40,174] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:48:40,208] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:48:40,306] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:48:40,373] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:48:40,381] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:48:40,384] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:48:54,493] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:48:54,849] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:48:55,291] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:48:55,401] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:48:55,601] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:48:55,702] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:48:55,869] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:48:56,197] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:48:56,411] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:48:56,488] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:48:56,570] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:48:56,861] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:48:57,012] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:48:57,065] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:48:57,452] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:48:57,485] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:49:09,616] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:49:11,238] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:49:11,806] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:49:11,809] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:49:11,881] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:49:11,932] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:49:12,157] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:49:12,534] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:49:12,557] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:49:12,751] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:49:12,783] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:49:13,060] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:49:13,116] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:49:13,343] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:49:13,730] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:49:13,823] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:49:24,776] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:49:26,697] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:49:27,464] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:49:27,523] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:49:27,936] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:49:28,080] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:49:28,134] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:49:28,503] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:49:28,526] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:49:28,772] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:49:29,223] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:49:29,243] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:49:29,312] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:49:29,460] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:49:29,717] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:49:29,754] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:49:42,116] [INFO] [logging.py:129:log_dist] [Rank 0] [Torch] Checkpoint 15 is about to be saved! [2024-12-05 22:49:42,148] [INFO] [logging.py:129:log_dist] [Rank 0] Saving model checkpoint: exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/mp_rank_00_model_states.pt [2024-12-05 22:49:42,148] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/mp_rank_00_model_states.pt... [2024-12-05 22:49:44,001] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/mp_rank_00_model_states.pt. [2024-12-05 22:49:44,160] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt... [2024-12-05 22:49:44,161] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt... [2024-12-05 22:49:44,162] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt... [2024-12-05 22:49:44,142] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_12_mp_rank_00_optim_states.pt... [2024-12-05 22:49:44,142] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_14_mp_rank_00_optim_states.pt... [2024-12-05 22:49:44,163] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt... [2024-12-05 22:49:44,143] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_13_mp_rank_00_optim_states.pt... [2024-12-05 22:49:44,143] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_10_mp_rank_00_optim_states.pt... [2024-12-05 22:49:44,143] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_15_mp_rank_00_optim_states.pt... [2024-12-05 22:49:44,143] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_11_mp_rank_00_optim_states.pt... [2024-12-05 22:49:44,144] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_9_mp_rank_00_optim_states.pt... [2024-12-05 22:49:44,165] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt... [2024-12-05 22:49:44,165] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt... [2024-12-05 22:49:44,145] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_8_mp_rank_00_optim_states.pt... [2024-12-05 22:49:44,166] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt... [2024-12-05 22:49:44,166] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt... [2024-12-05 22:49:44,770] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt. [2024-12-05 22:49:44,770] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt [2024-12-05 22:49:44,770] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 15 is ready now! [2024-12-05 22:49:44,891] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt. [2024-12-05 22:49:44,896] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt [2024-12-05 22:49:44,896] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 15 is ready now! [2024-12-05 22:49:44,896] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt. [2024-12-05 22:49:44,897] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt [2024-12-05 22:49:44,897] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 15 is ready now! [2024-12-05 22:49:44,882] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_11_mp_rank_00_optim_states.pt. [2024-12-05 22:49:44,882] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_11_mp_rank_00_optim_states.pt [2024-12-05 22:49:44,883] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 15 is ready now! [2024-12-05 22:49:44,891] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_9_mp_rank_00_optim_states.pt. [2024-12-05 22:49:44,891] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_9_mp_rank_00_optim_states.pt [2024-12-05 22:49:44,891] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 15 is ready now! [2024-12-05 22:49:44,926] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt. [2024-12-05 22:49:44,926] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt [2024-12-05 22:49:44,926] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 15 is ready now! [2024-12-05 22:49:44,927] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt. [2024-12-05 22:49:44,928] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt [2024-12-05 22:49:44,928] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 15 is ready now! [2024-12-05 22:49:44,940] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt. [2024-12-05 22:49:44,941] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt [2024-12-05 22:49:44,941] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 15 is ready now! [2024-12-05 22:49:44,944] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_12_mp_rank_00_optim_states.pt. [2024-12-05 22:49:44,944] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_12_mp_rank_00_optim_states.pt [2024-12-05 22:49:44,944] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 15 is ready now! [2024-12-05 22:49:44,948] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_8_mp_rank_00_optim_states.pt. [2024-12-05 22:49:44,949] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_8_mp_rank_00_optim_states.pt [2024-12-05 22:49:44,949] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 15 is ready now! [2024-12-05 22:49:44,957] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_14_mp_rank_00_optim_states.pt. [2024-12-05 22:49:44,957] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_14_mp_rank_00_optim_states.pt [2024-12-05 22:49:44,957] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 15 is ready now! [2024-12-05 22:49:44,960] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_10_mp_rank_00_optim_states.pt. [2024-12-05 22:49:44,960] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_10_mp_rank_00_optim_states.pt [2024-12-05 22:49:44,960] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 15 is ready now! [2024-12-05 22:49:44,986] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt. [2024-12-05 22:49:44,986] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt [2024-12-05 22:49:44,986] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 15 is ready now! [2024-12-05 22:49:45,010] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt. [2024-12-05 22:49:45,010] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt [2024-12-05 22:49:45,010] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 15 is ready now! [2024-12-05 22:49:45,320] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_13_mp_rank_00_optim_states.pt. [2024-12-05 22:49:45,320] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_13_mp_rank_00_optim_states.pt [2024-12-05 22:49:45,320] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 15 is ready now! [2024-12-05 22:49:45,324] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_15_mp_rank_00_optim_states.pt. [2024-12-05 22:49:45,324] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_15/15/bf16_zero_pp_rank_15_mp_rank_00_optim_states.pt [2024-12-05 22:49:45,324] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 15 is ready now! [cnode7-012:0/16] 2024-12-05 22:49:45,352 (deepspeed_trainer:158) INFO: 15epoch results: [train] iter_time=0.130, loss_ctc=79.345, loss_att=62.023, acc=0.684, loss=67.221, grad_norm=4.221, loss_scale=1.000, learning_rate=1.051e-04, step_time=0.366, time=2 hours, 4 minutes and 6.36 seconds, total_count=225015, gpu_max_cached_mem_GB=36.545, [valid] loss_ctc=6.031, cer_ctc=0.182, loss_att=8.125, acc=0.792, cer=0.321, wer=1.000, loss=7.500, time=1 minute and 9.18 seconds, total_count=15, gpu_max_cached_mem_GB=36.545 [cnode7-012:0/16] 2024-12-05 22:49:48,166 (multiple_iter_factory:32) INFO: Building 0th iter-factory... [cnode7-012:0/16] 2024-12-05 22:50:15,459 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-05 22:50:30,449 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.4", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.4", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.4", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.4", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-05 22:50:30,450 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.4, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-05 22:50:30,452 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-05 22:50:50,000] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:50:51,260] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:50:51,479] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:50:51,773] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:50:51,824] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:50:52,068] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:50:52,228] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:50:52,603] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:50:52,847] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:50:52,895] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:50:53,298] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:50:53,311] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:50:53,533] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:50:53,591] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:50:53,885] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:50:53,987] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:51:38,441] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:51:38,816] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:51:39,013] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:51:39,507] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:51:39,895] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:51:40,077] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:51:40,145] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:51:40,983] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:51:41,244] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:51:41,669] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:51:41,745] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:51:41,784] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:51:41,958] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:51:42,259] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:51:44,544] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:51:44,867] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:52:25,730] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:52:26,043] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:52:26,169] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:52:26,679] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:52:27,001] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:52:27,566] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:52:28,956] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:52:29,353] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:52:29,509] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:52:29,938] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:52:30,008] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:52:30,017] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:52:31,007] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:52:32,650] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:52:33,832] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:52:34,231] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:53:13,966] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:53:14,003] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:53:14,104] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:53:14,304] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:53:15,190] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:53:15,714] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:53:16,723] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:53:17,517] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:53:17,578] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:53:17,684] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:53:18,176] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:53:18,457] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:53:19,463] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:53:20,976] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:53:22,739] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 22:53:24,213] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-05 22:54:25,725 (deepspeed_trainer:228) INFO: 16epoch:train:1-100batch: iter_time=2.414, loss_ctc=76.526, loss_att=61.424, acc=0.686, loss=65.947, grad_norm=4.294, loss_scale=1.000, learning_rate=1.033e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 22:55:02,084 (deepspeed_trainer:228) INFO: 16epoch:train:101-200batch: iter_time=1.160e-04, loss_ctc=83.031, loss_att=69.334, acc=0.675, loss=73.442, grad_norm=4.580, loss_scale=1.000, learning_rate=1.032e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-05 22:55:37,422 (deepspeed_trainer:228) INFO: 16epoch:train:201-300batch: iter_time=1.116e-04, loss_ctc=74.623, loss_att=53.913, acc=0.687, loss=60.109, grad_norm=4.407, loss_scale=1.000, learning_rate=1.032e-04, step_time=0.353 [cnode7-012:0/16] 2024-12-05 22:56:13,409 (deepspeed_trainer:228) INFO: 16epoch:train:301-400batch: iter_time=1.103e-04, loss_ctc=92.089, loss_att=73.488, acc=0.668, loss=79.086, grad_norm=5.333, loss_scale=1.000, learning_rate=1.032e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 22:56:49,207 (deepspeed_trainer:228) INFO: 16epoch:train:401-500batch: iter_time=1.133e-04, loss_ctc=82.631, loss_att=60.733, acc=0.694, loss=67.331, grad_norm=4.761, loss_scale=1.000, learning_rate=1.032e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-05 22:57:24,558 (deepspeed_trainer:228) INFO: 16epoch:train:501-600batch: iter_time=1.146e-04, loss_ctc=67.978, loss_att=49.604, acc=0.692, loss=55.097, grad_norm=3.760, loss_scale=1.000, learning_rate=1.032e-04, step_time=0.354 [cnode7-012:0/16] 2024-12-05 22:58:00,238 (deepspeed_trainer:228) INFO: 16epoch:train:601-700batch: iter_time=1.144e-04, loss_ctc=79.243, loss_att=62.819, acc=0.673, loss=67.750, grad_norm=4.584, loss_scale=1.000, learning_rate=1.031e-04, step_time=0.356 [cnode7-012:0/16] 2024-12-05 22:58:36,108 (deepspeed_trainer:228) INFO: 16epoch:train:701-800batch: iter_time=1.128e-04, loss_ctc=83.038, loss_att=66.172, acc=0.680, loss=71.213, grad_norm=5.244, loss_scale=1.000, learning_rate=1.031e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 22:59:12,015 (deepspeed_trainer:228) INFO: 16epoch:train:801-900batch: iter_time=1.146e-04, loss_ctc=81.324, loss_att=62.654, acc=0.679, loss=68.232, grad_norm=4.526, loss_scale=1.000, learning_rate=1.031e-04, step_time=0.359 [2024-12-05 22:59:47,868] [INFO] [logging.py:129:log_dist] [Rank 0] step=226000, skipped=0, lr=[np.float64(0.0001030505799981523)], mom=[[0.9, 0.98]] [2024-12-05 22:59:47,869] [INFO] [timer.py:264:stop] epoch=0/micro_step=46000/global_step=46000, RunningAvgSamplesPerSec=45.10319913969565, CurrSamplesPerSec=45.730771326793004, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 22:59:47,871 (deepspeed_trainer:228) INFO: 16epoch:train:901-1000batch: iter_time=1.147e-04, loss_ctc=73.036, loss_att=57.578, acc=0.697, loss=62.184, grad_norm=4.192, loss_scale=1.000, learning_rate=1.031e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 23:00:23,771 (deepspeed_trainer:228) INFO: 16epoch:train:1001-1100batch: iter_time=1.147e-04, loss_ctc=80.194, loss_att=59.758, acc=0.697, loss=65.889, grad_norm=4.435, loss_scale=1.000, learning_rate=1.030e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 23:00:59,784 (deepspeed_trainer:228) INFO: 16epoch:train:1101-1200batch: iter_time=1.109e-04, loss_ctc=79.349, loss_att=62.099, acc=0.675, loss=67.252, grad_norm=4.494, loss_scale=1.000, learning_rate=1.030e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 23:01:35,869 (deepspeed_trainer:228) INFO: 16epoch:train:1201-1300batch: iter_time=1.116e-04, loss_ctc=75.445, loss_att=58.568, acc=0.685, loss=63.638, grad_norm=4.284, loss_scale=1.000, learning_rate=1.030e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 23:02:11,658 (deepspeed_trainer:228) INFO: 16epoch:train:1301-1400batch: iter_time=1.133e-04, loss_ctc=70.224, loss_att=61.431, acc=0.677, loss=64.053, grad_norm=4.043, loss_scale=1.000, learning_rate=1.030e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-05 23:02:47,561 (deepspeed_trainer:228) INFO: 16epoch:train:1401-1500batch: iter_time=1.172e-04, loss_ctc=73.843, loss_att=58.911, acc=0.686, loss=63.408, grad_norm=4.131, loss_scale=1.000, learning_rate=1.029e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 23:03:23,540 (deepspeed_trainer:228) INFO: 16epoch:train:1501-1600batch: iter_time=1.178e-04, loss_ctc=79.721, loss_att=57.179, acc=0.682, loss=63.925, grad_norm=4.940, loss_scale=1.000, learning_rate=1.029e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 23:03:59,555 (deepspeed_trainer:228) INFO: 16epoch:train:1601-1700batch: iter_time=1.155e-04, loss_ctc=76.647, loss_att=62.397, acc=0.692, loss=66.678, grad_norm=4.107, loss_scale=1.000, learning_rate=1.029e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 23:04:35,549 (deepspeed_trainer:228) INFO: 16epoch:train:1701-1800batch: iter_time=1.144e-04, loss_ctc=78.679, loss_att=59.334, acc=0.704, loss=65.162, grad_norm=4.168, loss_scale=1.000, learning_rate=1.029e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 23:05:05,406 (multiple_iter_factory:32) INFO: Building 1th iter-factory... [cnode7-012:0/16] 2024-12-05 23:05:31,584 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-05 23:05:48,542 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.2", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.2", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.2", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.2", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-05 23:05:48,542 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.2, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-05 23:05:48,544 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-05 23:06:16,158] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:06:16,184] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:06:16,418] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:06:17,916] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:06:18,555] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:06:18,580] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:06:18,853] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:06:20,201] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:06:24,415] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:06:24,478] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:06:24,501] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:06:24,616] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:06:24,664] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:06:24,725] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:06:24,790] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:06:24,824] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:07:05,664] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:07:06,874] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:07:08,336] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:07:08,967] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:07:10,487] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:07:10,772] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:07:10,963] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:07:11,355] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:07:17,477] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:07:17,723] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:07:18,130] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:07:19,001] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:07:19,092] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:07:19,471] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:07:19,512] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:07:19,554] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:07:53,460] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:07:56,306] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:07:58,904] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:08:00,727] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:08:01,280] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:08:01,346] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:08:01,747] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:08:02,069] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:08:06,345] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:08:06,784] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:08:07,159] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:08:07,552] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:08:07,589] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:08:08,389] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:08:08,823] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:08:08,998] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:08:41,277] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:08:44,955] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:08:48,609] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:08:50,556] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:08:51,188] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:08:51,652] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:08:52,234] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:08:52,297] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:08:54,229] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:08:55,325] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:08:56,303] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:08:56,369] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:08:56,751] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:08:57,036] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:08:57,534] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:08:57,614] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-05 23:09:34,046 (deepspeed_trainer:228) INFO: 16epoch:train:1801-1900batch: iter_time=2.486, loss_ctc=86.281, loss_att=68.554, acc=0.693, loss=73.884, grad_norm=4.283, loss_scale=1.000, learning_rate=1.029e-04, step_time=0.499 [2024-12-05 23:10:10,360] [INFO] [logging.py:129:log_dist] [Rank 0] step=227000, skipped=0, lr=[np.float64(0.00010282334680975977)], mom=[[0.9, 0.98]] [2024-12-05 23:10:10,360] [INFO] [timer.py:264:stop] epoch=0/micro_step=47000/global_step=47000, RunningAvgSamplesPerSec=45.07770843809536, CurrSamplesPerSec=47.709401158860246, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 23:10:10,362 (deepspeed_trainer:228) INFO: 16epoch:train:1901-2000batch: iter_time=1.074e-04, loss_ctc=77.411, loss_att=61.094, acc=0.687, loss=66.004, grad_norm=4.338, loss_scale=1.000, learning_rate=1.028e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-05 23:10:46,755 (deepspeed_trainer:228) INFO: 16epoch:train:2001-2100batch: iter_time=1.104e-04, loss_ctc=77.514, loss_att=61.215, acc=0.689, loss=66.102, grad_norm=4.677, loss_scale=1.000, learning_rate=1.028e-04, step_time=0.364 [cnode7-012:0/16] 2024-12-05 23:11:23,140 (deepspeed_trainer:228) INFO: 16epoch:train:2101-2200batch: iter_time=1.098e-04, loss_ctc=84.199, loss_att=65.028, acc=0.677, loss=70.753, grad_norm=4.883, loss_scale=1.000, learning_rate=1.028e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-05 23:11:59,229 (deepspeed_trainer:228) INFO: 16epoch:train:2201-2300batch: iter_time=1.124e-04, loss_ctc=77.626, loss_att=58.758, acc=0.685, loss=64.381, grad_norm=4.476, loss_scale=1.000, learning_rate=1.028e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 23:12:35,488 (deepspeed_trainer:228) INFO: 16epoch:train:2301-2400batch: iter_time=1.103e-04, loss_ctc=86.732, loss_att=67.661, acc=0.682, loss=73.368, grad_norm=4.422, loss_scale=1.000, learning_rate=1.027e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 23:13:11,283 (deepspeed_trainer:228) INFO: 16epoch:train:2401-2500batch: iter_time=1.111e-04, loss_ctc=64.399, loss_att=46.455, acc=0.700, loss=51.833, grad_norm=3.588, loss_scale=1.000, learning_rate=1.027e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-05 23:13:48,167 (deepspeed_trainer:228) INFO: 16epoch:train:2501-2600batch: iter_time=1.109e-04, loss_ctc=81.212, loss_att=65.449, acc=0.683, loss=70.176, grad_norm=4.485, loss_scale=1.000, learning_rate=1.027e-04, step_time=0.368 [cnode7-012:0/16] 2024-12-05 23:14:24,647 (deepspeed_trainer:228) INFO: 16epoch:train:2601-2700batch: iter_time=1.194e-04, loss_ctc=76.002, loss_att=61.612, acc=0.683, loss=65.959, grad_norm=4.762, loss_scale=1.000, learning_rate=1.027e-04, step_time=0.365 [cnode7-012:0/16] 2024-12-05 23:15:00,490 (deepspeed_trainer:228) INFO: 16epoch:train:2701-2800batch: iter_time=1.112e-04, loss_ctc=84.458, loss_att=62.874, acc=0.681, loss=69.351, grad_norm=4.333, loss_scale=1.000, learning_rate=1.027e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 23:15:36,289 (deepspeed_trainer:228) INFO: 16epoch:train:2801-2900batch: iter_time=1.129e-04, loss_ctc=72.548, loss_att=58.348, acc=0.703, loss=62.640, grad_norm=3.775, loss_scale=1.000, learning_rate=1.026e-04, step_time=0.358 [2024-12-05 23:16:12,094] [INFO] [logging.py:129:log_dist] [Rank 0] step=228000, skipped=0, lr=[np.float64(0.00010259761021400288)], mom=[[0.9, 0.98]] [2024-12-05 23:16:12,094] [INFO] [timer.py:264:stop] epoch=0/micro_step=48000/global_step=48000, RunningAvgSamplesPerSec=45.086437711023486, CurrSamplesPerSec=47.79717208615908, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 23:16:12,096 (deepspeed_trainer:228) INFO: 16epoch:train:2901-3000batch: iter_time=1.128e-04, loss_ctc=79.968, loss_att=60.499, acc=0.695, loss=66.358, grad_norm=4.317, loss_scale=1.000, learning_rate=1.026e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 23:16:47,528 (deepspeed_trainer:228) INFO: 16epoch:train:3001-3100batch: iter_time=1.105e-04, loss_ctc=73.772, loss_att=54.928, acc=0.690, loss=60.578, grad_norm=4.471, loss_scale=1.000, learning_rate=1.026e-04, step_time=0.354 [cnode7-012:0/16] 2024-12-05 23:17:23,369 (deepspeed_trainer:228) INFO: 16epoch:train:3101-3200batch: iter_time=1.097e-04, loss_ctc=74.369, loss_att=58.791, acc=0.679, loss=63.436, grad_norm=4.435, loss_scale=1.000, learning_rate=1.026e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 23:17:59,181 (deepspeed_trainer:228) INFO: 16epoch:train:3201-3300batch: iter_time=1.116e-04, loss_ctc=71.273, loss_att=61.042, acc=0.680, loss=64.089, grad_norm=4.256, loss_scale=1.000, learning_rate=1.025e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-05 23:18:35,135 (deepspeed_trainer:228) INFO: 16epoch:train:3301-3400batch: iter_time=1.118e-04, loss_ctc=76.429, loss_att=60.061, acc=0.686, loss=64.976, grad_norm=4.315, loss_scale=1.000, learning_rate=1.025e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 23:19:10,720 (deepspeed_trainer:228) INFO: 16epoch:train:3401-3500batch: iter_time=1.107e-04, loss_ctc=75.029, loss_att=55.137, acc=0.689, loss=61.133, grad_norm=4.558, loss_scale=1.000, learning_rate=1.025e-04, step_time=0.355 [cnode7-012:0/16] 2024-12-05 23:19:46,532 (deepspeed_trainer:228) INFO: 16epoch:train:3501-3600batch: iter_time=1.117e-04, loss_ctc=78.749, loss_att=61.462, acc=0.699, loss=66.641, grad_norm=4.250, loss_scale=1.000, learning_rate=1.025e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 23:20:22,667 (deepspeed_trainer:228) INFO: 16epoch:train:3601-3700batch: iter_time=1.104e-04, loss_ctc=80.267, loss_att=64.486, acc=0.710, loss=69.239, grad_norm=3.957, loss_scale=1.000, learning_rate=1.025e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 23:20:43,314 (multiple_iter_factory:32) INFO: Building 2th iter-factory... [cnode7-012:0/16] 2024-12-05 23:21:10,391 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-05 23:21:26,813 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.5", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.5", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.5", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.5", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-05 23:21:26,814 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.5, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-05 23:21:26,816 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-05 23:21:51,686] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:21:52,291] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:21:54,456] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:21:54,596] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:21:54,742] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:21:55,301] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:21:55,492] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:21:55,620] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:21:56,337] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:21:56,350] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:21:56,425] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:21:57,775] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:21:57,849] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:21:57,863] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:21:58,384] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:21:58,462] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:22:39,989] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:22:40,720] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:22:42,489] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:22:42,582] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:22:43,546] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:22:44,047] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:22:44,251] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:22:44,684] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:22:44,758] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:22:45,210] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:22:45,379] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:22:47,034] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:22:47,296] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:22:47,840] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:22:48,062] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:22:48,109] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:23:26,859] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:23:27,823] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:23:31,313] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:23:30,959] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:23:32,438] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:23:32,635] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:23:33,583] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:23:33,851] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:23:34,581] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:23:34,721] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:23:36,962] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:23:36,792] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:23:37,277] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:23:37,626] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:23:37,923] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:23:38,348] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:24:14,985] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:24:15,327] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:24:18,115] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:24:18,863] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:24:19,922] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:24:20,928] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:24:21,004] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:24:21,395] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:24:23,079] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:24:25,238] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:24:26,467] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:24:26,095] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:24:26,911] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:24:27,291] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:24:28,084] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:24:30,849] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-05 23:25:14,597 (deepspeed_trainer:228) INFO: 16epoch:train:3701-3800batch: iter_time=2.443, loss_ctc=78.216, loss_att=60.042, acc=0.688, loss=65.540, grad_norm=4.385, loss_scale=1.000, learning_rate=1.024e-04, step_time=0.476 [cnode7-012:0/16] 2024-12-05 23:25:50,650 (deepspeed_trainer:228) INFO: 16epoch:train:3801-3900batch: iter_time=1.087e-04, loss_ctc=83.302, loss_att=66.664, acc=0.683, loss=71.637, grad_norm=4.505, loss_scale=1.000, learning_rate=1.024e-04, step_time=0.360 [2024-12-05 23:26:26,828] [INFO] [logging.py:129:log_dist] [Rank 0] step=229000, skipped=0, lr=[np.float64(0.00010237335385470359)], mom=[[0.9, 0.98]] [2024-12-05 23:26:26,829] [INFO] [timer.py:264:stop] epoch=0/micro_step=49000/global_step=49000, RunningAvgSamplesPerSec=45.07221481851143, CurrSamplesPerSec=46.38911180817628, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 23:26:26,831 (deepspeed_trainer:228) INFO: 16epoch:train:3901-4000batch: iter_time=1.199e-04, loss_ctc=73.464, loss_att=57.603, acc=0.690, loss=62.351, grad_norm=4.057, loss_scale=1.000, learning_rate=1.024e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 23:27:02,767 (deepspeed_trainer:228) INFO: 16epoch:train:4001-4100batch: iter_time=1.111e-04, loss_ctc=87.557, loss_att=70.241, acc=0.675, loss=75.419, grad_norm=4.898, loss_scale=1.000, learning_rate=1.024e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 23:27:38,679 (deepspeed_trainer:228) INFO: 16epoch:train:4101-4200batch: iter_time=1.108e-04, loss_ctc=76.604, loss_att=60.895, acc=0.672, loss=65.629, grad_norm=4.836, loss_scale=1.000, learning_rate=1.023e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 23:28:14,859 (deepspeed_trainer:228) INFO: 16epoch:train:4201-4300batch: iter_time=1.092e-04, loss_ctc=83.176, loss_att=62.387, acc=0.688, loss=68.641, grad_norm=4.859, loss_scale=1.000, learning_rate=1.023e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 23:28:50,532 (deepspeed_trainer:228) INFO: 16epoch:train:4301-4400batch: iter_time=1.175e-04, loss_ctc=68.481, loss_att=51.758, acc=0.683, loss=56.797, grad_norm=4.117, loss_scale=1.000, learning_rate=1.023e-04, step_time=0.356 [cnode7-012:0/16] 2024-12-05 23:29:26,540 (deepspeed_trainer:228) INFO: 16epoch:train:4401-4500batch: iter_time=1.228e-04, loss_ctc=79.067, loss_att=65.306, acc=0.685, loss=69.448, grad_norm=4.517, loss_scale=1.000, learning_rate=1.023e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 23:30:02,630 (deepspeed_trainer:228) INFO: 16epoch:train:4501-4600batch: iter_time=1.219e-04, loss_ctc=73.749, loss_att=59.771, acc=0.682, loss=63.976, grad_norm=4.222, loss_scale=1.000, learning_rate=1.023e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 23:30:38,520 (deepspeed_trainer:228) INFO: 16epoch:train:4601-4700batch: iter_time=1.178e-04, loss_ctc=77.567, loss_att=57.024, acc=0.676, loss=63.165, grad_norm=4.411, loss_scale=1.000, learning_rate=1.022e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-05 23:31:14,509 (deepspeed_trainer:228) INFO: 16epoch:train:4701-4800batch: iter_time=1.077e-04, loss_ctc=79.551, loss_att=62.931, acc=0.703, loss=67.921, grad_norm=4.161, loss_scale=1.000, learning_rate=1.022e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-05 23:31:50,551 (deepspeed_trainer:228) INFO: 16epoch:train:4801-4900batch: iter_time=1.146e-04, loss_ctc=75.036, loss_att=57.065, acc=0.691, loss=62.463, grad_norm=4.450, loss_scale=1.000, learning_rate=1.022e-04, step_time=0.360 [2024-12-05 23:32:26,498] [INFO] [logging.py:129:log_dist] [Rank 0] step=230000, skipped=0, lr=[np.float64(0.00010215056162485288)], mom=[[0.9, 0.98]] [2024-12-05 23:32:26,498] [INFO] [timer.py:264:stop] epoch=0/micro_step=50000/global_step=50000, RunningAvgSamplesPerSec=45.084841899796864, CurrSamplesPerSec=47.88553990112196, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 23:32:26,500 (deepspeed_trainer:228) INFO: 16epoch:train:4901-5000batch: iter_time=1.218e-04, loss_ctc=77.760, loss_att=57.341, acc=0.680, loss=63.456, grad_norm=5.419, loss_scale=1.000, learning_rate=1.022e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 23:33:02,844 (deepspeed_trainer:228) INFO: 16epoch:train:5001-5100batch: iter_time=1.254e-04, loss_ctc=70.434, loss_att=58.010, acc=0.672, loss=61.752, grad_norm=4.627, loss_scale=1.000, learning_rate=1.021e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-05 23:33:39,179 (deepspeed_trainer:228) INFO: 16epoch:train:5101-5200batch: iter_time=1.320e-04, loss_ctc=70.031, loss_att=58.743, acc=0.679, loss=62.171, grad_norm=4.093, loss_scale=1.000, learning_rate=1.021e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-05 23:34:15,554 (deepspeed_trainer:228) INFO: 16epoch:train:5201-5300batch: iter_time=1.230e-04, loss_ctc=80.761, loss_att=61.291, acc=0.682, loss=67.094, grad_norm=4.572, loss_scale=1.000, learning_rate=1.021e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-05 23:34:51,769 (deepspeed_trainer:228) INFO: 16epoch:train:5301-5400batch: iter_time=1.231e-04, loss_ctc=71.889, loss_att=56.318, acc=0.690, loss=60.983, grad_norm=4.088, loss_scale=1.000, learning_rate=1.021e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 23:35:27,925 (deepspeed_trainer:228) INFO: 16epoch:train:5401-5500batch: iter_time=1.194e-04, loss_ctc=81.575, loss_att=61.341, acc=0.702, loss=67.431, grad_norm=4.330, loss_scale=1.000, learning_rate=1.021e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 23:36:03,723 (deepspeed_trainer:228) INFO: 16epoch:train:5501-5600batch: iter_time=1.153e-04, loss_ctc=77.709, loss_att=61.787, acc=0.696, loss=66.546, grad_norm=4.111, loss_scale=1.000, learning_rate=1.020e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-05 23:36:15,582 (multiple_iter_factory:32) INFO: Building 3th iter-factory... [cnode7-012:0/16] 2024-12-05 23:36:41,857 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-05 23:36:58,227 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.6", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.6", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.6", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.6", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-05 23:36:58,227 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.6, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-05 23:36:58,229 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-05 23:37:24,424] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:37:26,424] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:37:26,063] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:37:26,471] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:37:27,514] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:37:27,947] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:37:28,508] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:37:28,555] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:37:28,262] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:37:28,712] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:37:28,797] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:37:28,950] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:37:28,648] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:37:29,280] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:37:28,947] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:37:30,735] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:38:12,955] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:38:15,895] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:38:15,944] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:38:16,045] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:38:16,958] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:38:18,318] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:38:18,412] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:38:18,620] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:38:18,629] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:38:18,735] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:38:19,116] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:38:19,471] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:38:19,811] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:38:20,162] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:38:20,562] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:38:21,197] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:39:03,569] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:39:03,871] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:39:04,117] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:39:04,717] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:39:04,417] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:39:06,548] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:39:07,646] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:39:07,708] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:39:08,162] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:39:08,211] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:39:08,673] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:39:08,729] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:39:09,056] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:39:09,608] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:39:10,350] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:39:11,065] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:39:52,355] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:39:52,391] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:39:52,748] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:39:53,510] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:39:54,997] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:39:55,420] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:39:55,119] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:39:55,793] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:39:56,135] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:39:58,827] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:39:58,514] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:39:58,863] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:39:59,868] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:40:00,207] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:40:01,550] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:40:02,060] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-05 23:40:55,380 (deepspeed_trainer:228) INFO: 16epoch:train:5601-5700batch: iter_time=2.463, loss_ctc=78.638, loss_att=61.438, acc=0.696, loss=66.602, grad_norm=4.190, loss_scale=1.000, learning_rate=1.020e-04, step_time=0.453 [cnode7-012:0/16] 2024-12-05 23:41:31,675 (deepspeed_trainer:228) INFO: 16epoch:train:5701-5800batch: iter_time=1.101e-04, loss_ctc=82.477, loss_att=67.437, acc=0.674, loss=71.973, grad_norm=4.810, loss_scale=1.000, learning_rate=1.020e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-05 23:42:07,543 (deepspeed_trainer:228) INFO: 16epoch:train:5801-5900batch: iter_time=1.106e-04, loss_ctc=73.201, loss_att=57.294, acc=0.695, loss=62.086, grad_norm=4.166, loss_scale=1.000, learning_rate=1.020e-04, step_time=0.358 [2024-12-05 23:42:44,205] [INFO] [logging.py:129:log_dist] [Rank 0] step=231000, skipped=0, lr=[np.float64(0.00010192921766175164)], mom=[[0.9, 0.98]] [2024-12-05 23:42:44,206] [INFO] [timer.py:264:stop] epoch=0/micro_step=51000/global_step=51000, RunningAvgSamplesPerSec=45.0686744962813, CurrSamplesPerSec=46.995983717089565, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 23:42:44,208 (deepspeed_trainer:228) INFO: 16epoch:train:5901-6000batch: iter_time=1.133e-04, loss_ctc=87.881, loss_att=67.565, acc=0.677, loss=73.683, grad_norm=5.012, loss_scale=1.000, learning_rate=1.019e-04, step_time=0.366 [cnode7-012:0/16] 2024-12-05 23:43:20,532 (deepspeed_trainer:228) INFO: 16epoch:train:6001-6100batch: iter_time=1.131e-04, loss_ctc=71.771, loss_att=54.820, acc=0.699, loss=59.892, grad_norm=3.886, loss_scale=1.000, learning_rate=1.019e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-05 23:43:57,353 (deepspeed_trainer:228) INFO: 16epoch:train:6101-6200batch: iter_time=1.116e-04, loss_ctc=81.577, loss_att=61.172, acc=0.689, loss=67.294, grad_norm=4.188, loss_scale=1.000, learning_rate=1.019e-04, step_time=0.368 [cnode7-012:0/16] 2024-12-05 23:44:34,209 (deepspeed_trainer:228) INFO: 16epoch:train:6201-6300batch: iter_time=1.119e-04, loss_ctc=71.439, loss_att=53.722, acc=0.691, loss=59.051, grad_norm=4.096, loss_scale=1.000, learning_rate=1.019e-04, step_time=0.368 [cnode7-012:0/16] 2024-12-05 23:45:11,117 (deepspeed_trainer:228) INFO: 16epoch:train:6301-6400batch: iter_time=1.129e-04, loss_ctc=81.103, loss_att=67.674, acc=0.685, loss=71.711, grad_norm=4.903, loss_scale=1.000, learning_rate=1.019e-04, step_time=0.369 [cnode7-012:0/16] 2024-12-05 23:45:47,864 (deepspeed_trainer:228) INFO: 16epoch:train:6401-6500batch: iter_time=1.112e-04, loss_ctc=73.746, loss_att=59.048, acc=0.686, loss=63.450, grad_norm=4.105, loss_scale=1.000, learning_rate=1.018e-04, step_time=0.367 [cnode7-012:0/16] 2024-12-05 23:46:24,304 (deepspeed_trainer:228) INFO: 16epoch:train:6501-6600batch: iter_time=1.127e-04, loss_ctc=72.314, loss_att=55.620, acc=0.697, loss=60.638, grad_norm=3.981, loss_scale=1.000, learning_rate=1.018e-04, step_time=0.364 [cnode7-012:0/16] 2024-12-05 23:47:00,949 (deepspeed_trainer:228) INFO: 16epoch:train:6601-6700batch: iter_time=1.134e-04, loss_ctc=80.282, loss_att=62.040, acc=0.703, loss=67.531, grad_norm=4.218, loss_scale=1.000, learning_rate=1.018e-04, step_time=0.366 [cnode7-012:0/16] 2024-12-05 23:47:37,077 (deepspeed_trainer:228) INFO: 16epoch:train:6701-6800batch: iter_time=1.121e-04, loss_ctc=74.060, loss_att=55.992, acc=0.692, loss=61.419, grad_norm=4.217, loss_scale=1.000, learning_rate=1.018e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-05 23:48:13,135 (deepspeed_trainer:228) INFO: 16epoch:train:6801-6900batch: iter_time=1.136e-04, loss_ctc=78.843, loss_att=58.045, acc=0.687, loss=64.287, grad_norm=4.661, loss_scale=1.000, learning_rate=1.017e-04, step_time=0.360 [2024-12-05 23:48:49,417] [INFO] [logging.py:129:log_dist] [Rank 0] step=232000, skipped=0, lr=[np.float64(0.00010170930634226651)], mom=[[0.9, 0.98]] [2024-12-05 23:48:49,417] [INFO] [timer.py:264:stop] epoch=0/micro_step=52000/global_step=52000, RunningAvgSamplesPerSec=45.06832679997729, CurrSamplesPerSec=47.711538077782514, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 23:48:49,419 (deepspeed_trainer:228) INFO: 16epoch:train:6901-7000batch: iter_time=1.114e-04, loss_ctc=72.676, loss_att=63.379, acc=0.672, loss=66.188, grad_norm=4.045, loss_scale=1.000, learning_rate=1.017e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-05 23:49:25,201 (deepspeed_trainer:228) INFO: 16epoch:train:7001-7100batch: iter_time=1.113e-04, loss_ctc=70.527, loss_att=59.770, acc=0.686, loss=62.984, grad_norm=3.981, loss_scale=1.000, learning_rate=1.017e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-05 23:50:00,894 (deepspeed_trainer:228) INFO: 16epoch:train:7101-7200batch: iter_time=1.130e-04, loss_ctc=78.399, loss_att=55.913, acc=0.698, loss=62.682, grad_norm=4.296, loss_scale=1.000, learning_rate=1.017e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-05 23:50:36,580 (deepspeed_trainer:228) INFO: 16epoch:train:7201-7300batch: iter_time=1.106e-04, loss_ctc=73.301, loss_att=57.512, acc=0.696, loss=62.281, grad_norm=3.809, loss_scale=1.000, learning_rate=1.017e-04, step_time=0.356 [cnode7-012:0/16] 2024-12-05 23:51:12,515 (deepspeed_trainer:228) INFO: 16epoch:train:7301-7400batch: iter_time=1.126e-04, loss_ctc=78.739, loss_att=61.520, acc=0.707, loss=66.663, grad_norm=4.258, loss_scale=1.000, learning_rate=1.016e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 23:51:48,485 (deepspeed_trainer:228) INFO: 16epoch:train:7401-7500batch: iter_time=1.110e-04, loss_ctc=82.338, loss_att=62.011, acc=0.707, loss=68.136, grad_norm=4.339, loss_scale=1.000, learning_rate=1.016e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-05 23:51:50,929 (multiple_iter_factory:32) INFO: Building 4th iter-factory... [cnode7-012:0/16] 2024-12-05 23:52:18,011 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-05 23:52:33,264 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.7", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.7", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.7", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.7", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-05 23:52:33,265 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.7, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-05 23:52:33,267 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-05 23:53:00,584] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:53:01,242] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:53:02,694] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:53:03,131] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:53:03,154] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:53:02,845] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:53:03,351] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:53:03,890] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:53:04,287] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:53:04,324] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:53:04,668] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:53:04,461] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:53:04,858] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:53:04,999] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:53:05,591] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:53:05,660] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:53:49,111] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:53:49,651] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:53:51,026] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:53:51,738] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:53:51,756] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:53:52,156] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:53:52,368] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:53:52,615] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:53:52,861] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:53:53,158] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:53:53,482] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:53:53,863] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:53:54,513] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:53:54,553] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:53:55,362] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:53:56,038] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:54:36,782] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:54:37,702] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:54:38,634] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:54:39,044] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:54:39,955] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:54:40,217] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:54:40,452] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:54:41,125] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:54:41,572] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:54:41,939] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:54:43,052] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:54:43,485] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:54:44,002] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:54:44,760] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:54:44,976] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:54:45,977] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:55:24,707] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:55:26,272] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:55:26,817] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:55:27,614] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:55:27,767] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:55:28,367] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:55:28,744] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:55:28,810] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:55:29,491] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:55:29,899] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:55:32,023] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:55:32,694] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:55:32,742] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:55:33,193] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:55:33,824] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-05 23:55:35,536] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-05 23:56:37,778 (deepspeed_trainer:228) INFO: 16epoch:train:7501-7600batch: iter_time=2.438, loss_ctc=74.973, loss_att=60.652, acc=0.688, loss=64.933, grad_norm=4.207, loss_scale=1.000, learning_rate=1.016e-04, step_time=0.454 [cnode7-012:0/16] 2024-12-05 23:57:14,520 (deepspeed_trainer:228) INFO: 16epoch:train:7601-7700batch: iter_time=1.174e-04, loss_ctc=80.034, loss_att=67.823, acc=0.678, loss=71.473, grad_norm=4.668, loss_scale=1.000, learning_rate=1.016e-04, step_time=0.367 [cnode7-012:0/16] 2024-12-05 23:57:50,742 (deepspeed_trainer:228) INFO: 16epoch:train:7701-7800batch: iter_time=1.190e-04, loss_ctc=72.268, loss_att=51.375, acc=0.695, loss=57.655, grad_norm=4.354, loss_scale=1.000, learning_rate=1.015e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 23:58:27,610 (deepspeed_trainer:228) INFO: 16epoch:train:7801-7900batch: iter_time=1.159e-04, loss_ctc=87.879, loss_att=72.386, acc=0.668, loss=77.042, grad_norm=5.219, loss_scale=1.000, learning_rate=1.015e-04, step_time=0.368 [2024-12-05 23:59:03,929] [INFO] [logging.py:129:log_dist] [Rank 0] step=233000, skipped=0, lr=[np.float64(0.00010149081227819843)], mom=[[0.9, 0.98]] [2024-12-05 23:59:03,930] [INFO] [timer.py:264:stop] epoch=0/micro_step=53000/global_step=53000, RunningAvgSamplesPerSec=45.054298982029884, CurrSamplesPerSec=39.873879847718314, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-05 23:59:03,932 (deepspeed_trainer:228) INFO: 16epoch:train:7901-8000batch: iter_time=1.197e-04, loss_ctc=80.433, loss_att=60.690, acc=0.692, loss=66.607, grad_norm=4.642, loss_scale=1.000, learning_rate=1.015e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-05 23:59:39,708 (deepspeed_trainer:228) INFO: 16epoch:train:8001-8100batch: iter_time=1.215e-04, loss_ctc=66.275, loss_att=48.519, acc=0.695, loss=53.858, grad_norm=4.076, loss_scale=1.000, learning_rate=1.015e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-06 00:00:16,505 (deepspeed_trainer:228) INFO: 16epoch:train:8101-8200batch: iter_time=1.201e-04, loss_ctc=77.889, loss_att=61.551, acc=0.677, loss=66.440, grad_norm=4.249, loss_scale=1.000, learning_rate=1.015e-04, step_time=0.368 [cnode7-012:0/16] 2024-12-06 00:00:52,656 (deepspeed_trainer:228) INFO: 16epoch:train:8201-8300batch: iter_time=1.212e-04, loss_ctc=78.436, loss_att=63.356, acc=0.685, loss=67.857, grad_norm=4.739, loss_scale=1.000, learning_rate=1.014e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-06 00:01:29,003 (deepspeed_trainer:228) INFO: 16epoch:train:8301-8400batch: iter_time=1.169e-04, loss_ctc=79.578, loss_att=61.710, acc=0.675, loss=67.062, grad_norm=4.185, loss_scale=1.000, learning_rate=1.014e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-06 00:02:05,139 (deepspeed_trainer:228) INFO: 16epoch:train:8401-8500batch: iter_time=1.141e-04, loss_ctc=71.691, loss_att=55.629, acc=0.698, loss=60.420, grad_norm=3.985, loss_scale=1.000, learning_rate=1.014e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-06 00:02:41,531 (deepspeed_trainer:228) INFO: 16epoch:train:8501-8600batch: iter_time=1.155e-04, loss_ctc=78.558, loss_att=58.870, acc=0.701, loss=64.779, grad_norm=4.094, loss_scale=1.000, learning_rate=1.014e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-06 00:03:17,802 (deepspeed_trainer:228) INFO: 16epoch:train:8601-8700batch: iter_time=1.180e-04, loss_ctc=76.634, loss_att=60.017, acc=0.677, loss=65.011, grad_norm=4.606, loss_scale=1.000, learning_rate=1.013e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-06 00:03:54,482 (deepspeed_trainer:228) INFO: 16epoch:train:8701-8800batch: iter_time=1.215e-04, loss_ctc=73.705, loss_att=56.133, acc=0.688, loss=61.415, grad_norm=4.262, loss_scale=1.000, learning_rate=1.013e-04, step_time=0.366 [cnode7-012:0/16] 2024-12-06 00:04:30,842 (deepspeed_trainer:228) INFO: 16epoch:train:8801-8900batch: iter_time=1.149e-04, loss_ctc=69.522, loss_att=60.017, acc=0.676, loss=62.899, grad_norm=3.935, loss_scale=1.000, learning_rate=1.013e-04, step_time=0.363 [2024-12-06 00:05:07,178] [INFO] [logging.py:129:log_dist] [Rank 0] step=234000, skipped=0, lr=[np.float64(0.00010127372031175958)], mom=[[0.9, 0.98]] [2024-12-06 00:05:07,179] [INFO] [timer.py:264:stop] epoch=0/micro_step=54000/global_step=54000, RunningAvgSamplesPerSec=45.05833761215849, CurrSamplesPerSec=47.845664053252186, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-06 00:05:07,180 (deepspeed_trainer:228) INFO: 16epoch:train:8901-9000batch: iter_time=1.229e-04, loss_ctc=72.033, loss_att=58.097, acc=0.689, loss=62.288, grad_norm=3.748, loss_scale=1.000, learning_rate=1.013e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-06 00:05:43,530 (deepspeed_trainer:228) INFO: 16epoch:train:9001-9100batch: iter_time=1.217e-04, loss_ctc=77.197, loss_att=55.972, acc=0.685, loss=62.329, grad_norm=4.367, loss_scale=1.000, learning_rate=1.013e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-06 00:06:19,974 (deepspeed_trainer:228) INFO: 16epoch:train:9101-9200batch: iter_time=1.151e-04, loss_ctc=74.760, loss_att=60.903, acc=0.693, loss=65.104, grad_norm=4.181, loss_scale=1.000, learning_rate=1.012e-04, step_time=0.364 [cnode7-012:0/16] 2024-12-06 00:06:56,420 (deepspeed_trainer:228) INFO: 16epoch:train:9201-9300batch: iter_time=1.132e-04, loss_ctc=77.227, loss_att=58.269, acc=0.703, loss=63.955, grad_norm=3.990, loss_scale=1.000, learning_rate=1.012e-04, step_time=0.364 [cnode7-012:0/16] 2024-12-06 00:07:26,328 (multiple_iter_factory:32) INFO: Building 5th iter-factory... [cnode7-012:0/16] 2024-12-06 00:07:53,693 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-06 00:08:10,449 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.1", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.1", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.1", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.1", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-06 00:08:10,450 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.1, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-06 00:08:10,452 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-06 00:08:35,327] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:08:35,696] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:08:38,059] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:08:37,812] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:08:38,168] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:08:38,648] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:08:38,952] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:08:39,236] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:08:39,453] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:08:39,650] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:08:39,782] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:08:40,012] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:08:41,353] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:08:41,789] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:08:42,715] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:08:43,000] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:09:23,911] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:09:25,274] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:09:27,325] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:09:27,491] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:09:28,333] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:09:28,683] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:09:28,736] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:09:28,866] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:09:28,966] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:09:29,201] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:09:29,367] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:09:31,352] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:09:32,912] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:09:32,981] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:09:34,140] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:09:34,754] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:10:13,394] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:10:13,448] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:10:15,634] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:10:16,764] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:10:17,090] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:10:17,775] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:10:18,258] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:10:18,451] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:10:18,709] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:10:18,908] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:10:20,241] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:10:21,540] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:10:22,068] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:10:22,542] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:10:23,306] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:10:23,872] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:11:01,488] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:11:01,673] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:11:02,305] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:11:05,325] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:11:05,615] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:11:06,449] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:11:06,637] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:11:06,986] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:11:07,422] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:11:07,671] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:11:08,553] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:11:09,099] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:11:10,611] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:11:11,765] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:11:12,436] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:11:13,830] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-06 00:11:49,216 (deepspeed_trainer:228) INFO: 16epoch:train:9301-9400batch: iter_time=2.443, loss_ctc=84.402, loss_att=65.896, acc=0.697, loss=71.448, grad_norm=4.281, loss_scale=1.000, learning_rate=1.012e-04, step_time=0.484 [cnode7-012:0/16] 2024-12-06 00:12:25,807 (deepspeed_trainer:228) INFO: 16epoch:train:9401-9500batch: iter_time=1.074e-04, loss_ctc=76.009, loss_att=60.497, acc=0.688, loss=65.186, grad_norm=4.611, loss_scale=1.000, learning_rate=1.012e-04, step_time=0.366 [cnode7-012:0/16] 2024-12-06 00:13:01,985 (deepspeed_trainer:228) INFO: 16epoch:train:9501-9600batch: iter_time=1.355e-04, loss_ctc=75.198, loss_att=59.601, acc=0.692, loss=64.275, grad_norm=4.430, loss_scale=1.000, learning_rate=1.012e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-06 00:13:39,017 (deepspeed_trainer:228) INFO: 16epoch:train:9601-9700batch: iter_time=1.100e-04, loss_ctc=82.276, loss_att=63.011, acc=0.679, loss=68.779, grad_norm=5.245, loss_scale=1.000, learning_rate=1.011e-04, step_time=0.370 [cnode7-012:0/16] 2024-12-06 00:14:15,777 (deepspeed_trainer:228) INFO: 16epoch:train:9701-9800batch: iter_time=1.095e-04, loss_ctc=74.964, loss_att=58.928, acc=0.681, loss=63.745, grad_norm=4.453, loss_scale=1.000, learning_rate=1.011e-04, step_time=0.367 [cnode7-012:0/16] 2024-12-06 00:14:52,133 (deepspeed_trainer:228) INFO: 16epoch:train:9801-9900batch: iter_time=1.132e-04, loss_ctc=85.424, loss_att=66.241, acc=0.680, loss=71.969, grad_norm=4.867, loss_scale=1.000, learning_rate=1.011e-04, step_time=0.364 [2024-12-06 00:15:28,395] [INFO] [logging.py:129:log_dist] [Rank 0] step=235000, skipped=0, lr=[np.float64(0.00010105801551115674)], mom=[[0.9, 0.98]] [2024-12-06 00:15:28,395] [INFO] [timer.py:264:stop] epoch=0/micro_step=55000/global_step=55000, RunningAvgSamplesPerSec=45.03095265778918, CurrSamplesPerSec=40.37414624169425, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-06 00:15:28,397 (deepspeed_trainer:228) INFO: 16epoch:train:9901-10000batch: iter_time=1.154e-04, loss_ctc=63.271, loss_att=45.554, acc=0.704, loss=50.865, grad_norm=3.824, loss_scale=1.000, learning_rate=1.011e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-06 00:16:04,849 (deepspeed_trainer:228) INFO: 16epoch:train:10001-10100batch: iter_time=1.168e-04, loss_ctc=79.588, loss_att=64.346, acc=0.685, loss=68.902, grad_norm=4.240, loss_scale=1.000, learning_rate=1.010e-04, step_time=0.364 [cnode7-012:0/16] 2024-12-06 00:16:40,949 (deepspeed_trainer:228) INFO: 16epoch:train:10101-10200batch: iter_time=1.118e-04, loss_ctc=73.851, loss_att=59.114, acc=0.682, loss=63.534, grad_norm=4.561, loss_scale=1.000, learning_rate=1.010e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-06 00:17:17,185 (deepspeed_trainer:228) INFO: 16epoch:train:10201-10300batch: iter_time=1.114e-04, loss_ctc=82.931, loss_att=61.252, acc=0.682, loss=67.764, grad_norm=4.557, loss_scale=1.000, learning_rate=1.010e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-06 00:17:53,187 (deepspeed_trainer:228) INFO: 16epoch:train:10301-10400batch: iter_time=1.125e-04, loss_ctc=71.941, loss_att=57.348, acc=0.702, loss=61.707, grad_norm=4.095, loss_scale=1.000, learning_rate=1.010e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-06 00:18:29,525 (deepspeed_trainer:228) INFO: 16epoch:train:10401-10500batch: iter_time=1.149e-04, loss_ctc=78.480, loss_att=59.499, acc=0.698, loss=65.189, grad_norm=4.205, loss_scale=1.000, learning_rate=1.010e-04, step_time=0.363 [cnode7-012:0/16] 2024-12-06 00:19:05,357 (deepspeed_trainer:228) INFO: 16epoch:train:10501-10600batch: iter_time=1.138e-04, loss_ctc=72.527, loss_att=53.086, acc=0.691, loss=58.931, grad_norm=4.537, loss_scale=1.000, learning_rate=1.009e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-06 00:19:41,111 (deepspeed_trainer:228) INFO: 16epoch:train:10601-10700batch: iter_time=1.120e-04, loss_ctc=72.499, loss_att=57.376, acc=0.678, loss=61.936, grad_norm=4.116, loss_scale=1.000, learning_rate=1.009e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-06 00:20:17,121 (deepspeed_trainer:228) INFO: 16epoch:train:10701-10800batch: iter_time=1.132e-04, loss_ctc=71.526, loss_att=60.957, acc=0.674, loss=64.103, grad_norm=3.993, loss_scale=1.000, learning_rate=1.009e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-06 00:20:53,028 (deepspeed_trainer:228) INFO: 16epoch:train:10801-10900batch: iter_time=1.121e-04, loss_ctc=74.692, loss_att=58.922, acc=0.688, loss=63.669, grad_norm=3.976, loss_scale=1.000, learning_rate=1.009e-04, step_time=0.359 [2024-12-06 00:21:28,752] [INFO] [logging.py:129:log_dist] [Rank 0] step=236000, skipped=0, lr=[np.float64(0.00010084368316627737)], mom=[[0.9, 0.98]] [2024-12-06 00:21:28,753] [INFO] [timer.py:264:stop] epoch=0/micro_step=56000/global_step=56000, RunningAvgSamplesPerSec=45.042431877436734, CurrSamplesPerSec=45.46584154623431, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-06 00:21:28,755 (deepspeed_trainer:228) INFO: 16epoch:train:10901-11000batch: iter_time=1.115e-04, loss_ctc=72.466, loss_att=54.380, acc=0.690, loss=59.784, grad_norm=4.158, loss_scale=1.000, learning_rate=1.009e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-06 00:22:04,995 (deepspeed_trainer:228) INFO: 16epoch:train:11001-11100batch: iter_time=1.147e-04, loss_ctc=77.500, loss_att=60.801, acc=0.697, loss=65.798, grad_norm=3.753, loss_scale=1.000, learning_rate=1.008e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-06 00:22:41,007 (deepspeed_trainer:228) INFO: 16epoch:train:11101-11200batch: iter_time=1.126e-04, loss_ctc=79.771, loss_att=63.455, acc=0.706, loss=68.326, grad_norm=3.938, loss_scale=1.000, learning_rate=1.008e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-06 00:23:01,459 (multiple_iter_factory:32) INFO: Building 6th iter-factory... [cnode7-012:0/16] 2024-12-06 00:23:28,835 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-06 00:23:44,424 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.0", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.0", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.0", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.0", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-06 00:23:44,424 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.0, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-06 00:23:44,426 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-06 00:24:11,163] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:24:10,315] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:24:11,738] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:24:11,975] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:24:10,827] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:24:11,089] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:24:11,443] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:24:12,050] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:24:12,324] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:24:12,617] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:24:13,924] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:24:12,822] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:24:14,163] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:24:14,664] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:24:16,120] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:24:16,171] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:24:57,666] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:24:59,851] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:25:00,163] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:24:59,023] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:24:59,464] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:25:00,787] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:25:00,540] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:25:00,987] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:25:02,486] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:25:01,559] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:25:01,614] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:25:03,693] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:25:03,243] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:25:05,032] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:25:05,722] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:25:06,874] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:25:44,790] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:25:46,932] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:25:48,590] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:25:48,902] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:25:48,268] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:25:48,595] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:25:48,855] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:25:49,175] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:25:50,255] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:25:50,783] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:25:52,080] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:25:52,877] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:25:54,577] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:25:54,954] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:25:53,960] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:25:58,962] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:26:32,454] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:26:34,993] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:26:36,589] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:26:36,691] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:26:36,507] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:26:36,858] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:26:36,928] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:26:37,276] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:26:37,521] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:26:40,904] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:26:41,099] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:26:40,663] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:26:42,279] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:26:42,960] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:26:43,786] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:26:48,695] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-06 00:27:32,306 (deepspeed_trainer:228) INFO: 16epoch:train:11201-11300batch: iter_time=2.443, loss_ctc=77.172, loss_att=58.731, acc=0.696, loss=64.234, grad_norm=4.144, loss_scale=1.000, learning_rate=1.008e-04, step_time=0.470 [cnode7-012:0/16] 2024-12-06 00:28:08,924 (deepspeed_trainer:228) INFO: 16epoch:train:11301-11400batch: iter_time=1.069e-04, loss_ctc=81.612, loss_att=65.951, acc=0.689, loss=70.667, grad_norm=4.774, loss_scale=1.000, learning_rate=1.008e-04, step_time=0.365 [cnode7-012:0/16] 2024-12-06 00:28:45,050 (deepspeed_trainer:228) INFO: 16epoch:train:11401-11500batch: iter_time=1.085e-04, loss_ctc=72.498, loss_att=56.785, acc=0.698, loss=61.483, grad_norm=4.420, loss_scale=1.000, learning_rate=1.007e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-06 00:29:21,867 (deepspeed_trainer:228) INFO: 16epoch:train:11501-11600batch: iter_time=1.117e-04, loss_ctc=86.591, loss_att=70.226, acc=0.681, loss=75.081, grad_norm=5.082, loss_scale=1.000, learning_rate=1.007e-04, step_time=0.368 [cnode7-012:0/16] 2024-12-06 00:29:58,137 (deepspeed_trainer:228) INFO: 16epoch:train:11601-11700batch: iter_time=1.115e-04, loss_ctc=74.608, loss_att=58.292, acc=0.686, loss=63.193, grad_norm=4.766, loss_scale=1.000, learning_rate=1.007e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-06 00:30:34,387 (deepspeed_trainer:228) INFO: 16epoch:train:11701-11800batch: iter_time=1.114e-04, loss_ctc=82.081, loss_att=60.797, acc=0.701, loss=67.191, grad_norm=4.602, loss_scale=1.000, learning_rate=1.007e-04, step_time=0.362 [cnode7-012:0/16] 2024-12-06 00:31:09,996 (deepspeed_trainer:228) INFO: 16epoch:train:11801-11900batch: iter_time=1.106e-04, loss_ctc=67.597, loss_att=51.612, acc=0.692, loss=56.367, grad_norm=3.879, loss_scale=1.000, learning_rate=1.007e-04, step_time=0.356 [2024-12-06 00:31:45,893] [INFO] [logging.py:129:log_dist] [Rank 0] step=237000, skipped=0, lr=[np.float64(0.00010063070878447609)], mom=[[0.9, 0.98]] [2024-12-06 00:31:45,897] [INFO] [timer.py:264:stop] epoch=0/micro_step=57000/global_step=57000, RunningAvgSamplesPerSec=45.02539976244313, CurrSamplesPerSec=44.79098538306749, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-06 00:31:45,899 (deepspeed_trainer:228) INFO: 16epoch:train:11901-12000batch: iter_time=1.108e-04, loss_ctc=77.571, loss_att=63.073, acc=0.697, loss=67.451, grad_norm=4.686, loss_scale=1.000, learning_rate=1.006e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-06 00:32:21,904 (deepspeed_trainer:228) INFO: 16epoch:train:12001-12100batch: iter_time=1.109e-04, loss_ctc=72.284, loss_att=58.983, acc=0.693, loss=62.983, grad_norm=3.676, loss_scale=1.000, learning_rate=1.006e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-06 00:32:57,837 (deepspeed_trainer:228) INFO: 16epoch:train:12101-12200batch: iter_time=1.104e-04, loss_ctc=76.609, loss_att=56.371, acc=0.690, loss=62.441, grad_norm=4.140, loss_scale=1.000, learning_rate=1.006e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-06 00:33:33,557 (deepspeed_trainer:228) INFO: 16epoch:train:12201-12300batch: iter_time=1.090e-04, loss_ctc=78.674, loss_att=62.434, acc=0.710, loss=67.279, grad_norm=4.134, loss_scale=1.000, learning_rate=1.006e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-06 00:34:09,262 (deepspeed_trainer:228) INFO: 16epoch:train:12301-12400batch: iter_time=1.120e-04, loss_ctc=73.909, loss_att=56.817, acc=0.699, loss=61.917, grad_norm=3.961, loss_scale=1.000, learning_rate=1.006e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-06 00:34:45,425 (deepspeed_trainer:228) INFO: 16epoch:train:12401-12500batch: iter_time=1.118e-04, loss_ctc=76.323, loss_att=58.460, acc=0.688, loss=63.811, grad_norm=4.428, loss_scale=1.000, learning_rate=1.005e-04, step_time=0.361 [cnode7-012:0/16] 2024-12-06 00:35:21,351 (deepspeed_trainer:228) INFO: 16epoch:train:12501-12600batch: iter_time=1.116e-04, loss_ctc=70.231, loss_att=57.395, acc=0.684, loss=61.278, grad_norm=4.313, loss_scale=1.000, learning_rate=1.005e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-06 00:35:56,949 (deepspeed_trainer:228) INFO: 16epoch:train:12601-12700batch: iter_time=1.116e-04, loss_ctc=69.093, loss_att=58.258, acc=0.686, loss=61.518, grad_norm=3.970, loss_scale=1.000, learning_rate=1.005e-04, step_time=0.356 [cnode7-012:0/16] 2024-12-06 00:36:32,650 (deepspeed_trainer:228) INFO: 16epoch:train:12701-12800batch: iter_time=1.101e-04, loss_ctc=78.587, loss_att=59.705, acc=0.691, loss=65.347, grad_norm=4.328, loss_scale=1.000, learning_rate=1.005e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-06 00:37:08,465 (deepspeed_trainer:228) INFO: 16epoch:train:12801-12900batch: iter_time=1.089e-04, loss_ctc=71.390, loss_att=55.991, acc=0.697, loss=60.592, grad_norm=4.025, loss_scale=1.000, learning_rate=1.005e-04, step_time=0.358 [2024-12-06 00:37:44,195] [INFO] [logging.py:129:log_dist] [Rank 0] step=238000, skipped=0, lr=[np.float64(0.0001004190780864586)], mom=[[0.9, 0.98]] [2024-12-06 00:37:44,195] [INFO] [timer.py:264:stop] epoch=0/micro_step=58000/global_step=58000, RunningAvgSamplesPerSec=45.04086423541769, CurrSamplesPerSec=46.25329193334641, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-06 00:37:44,197 (deepspeed_trainer:228) INFO: 16epoch:train:12901-13000batch: iter_time=1.077e-04, loss_ctc=80.657, loss_att=61.227, acc=0.710, loss=67.056, grad_norm=4.133, loss_scale=1.000, learning_rate=1.004e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-06 00:38:19,917 (deepspeed_trainer:228) INFO: 16epoch:train:13001-13100batch: iter_time=1.091e-04, loss_ctc=77.315, loss_att=61.567, acc=0.708, loss=66.266, grad_norm=3.804, loss_scale=1.000, learning_rate=1.004e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-06 00:38:31,617 (multiple_iter_factory:32) INFO: Building 7th iter-factory... [cnode7-012:0/16] 2024-12-06 00:38:58,840 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-06 00:39:14,176 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.3", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.3", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.3", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.3", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-06 00:39:14,177 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.3, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-06 00:39:14,179 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-06 00:39:41,654] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:39:41,706] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:39:42,011] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:39:41,878] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:39:42,182] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:39:42,539] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:39:43,072] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:39:43,110] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:39:43,162] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:39:43,682] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:39:44,072] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:39:45,589] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:39:45,858] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:39:46,507] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:39:46,654] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:39:46,801] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:40:30,728] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:40:30,789] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:40:31,043] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:40:30,294] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:40:30,785] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:40:31,345] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:40:31,438] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:40:31,980] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:40:32,637] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:40:32,687] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:40:32,946] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:40:34,991] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:40:35,237] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:40:35,506] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:40:36,287] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:40:40,340] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:41:20,070] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:41:18,904] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:41:20,221] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:41:20,395] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:41:19,599] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:41:20,154] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:41:20,193] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:41:20,584] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:41:21,490] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:41:21,500] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:41:21,904] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:41:25,474] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:41:25,818] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:41:26,059] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:41:27,245] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:41:32,931] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:42:07,198] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:42:07,916] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:42:06,770] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:42:07,684] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:42:09,131] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:42:09,689] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:42:09,695] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:42:10,320] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:42:10,370] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:42:11,849] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:42:10,720] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:42:13,389] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:42:13,962] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:42:14,306] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:42:14,810] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:42:23,211] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-06 00:43:15,945 (deepspeed_trainer:228) INFO: 16epoch:train:13101-13200batch: iter_time=2.441, loss_ctc=78.123, loss_att=61.557, acc=0.693, loss=66.537, grad_norm=4.256, loss_scale=1.000, learning_rate=1.004e-04, step_time=0.519 [cnode7-012:0/16] 2024-12-06 00:43:52,406 (deepspeed_trainer:228) INFO: 16epoch:train:13201-13300batch: iter_time=1.101e-04, loss_ctc=81.757, loss_att=66.891, acc=0.675, loss=71.331, grad_norm=4.710, loss_scale=1.000, learning_rate=1.004e-04, step_time=0.364 [cnode7-012:0/16] 2024-12-06 00:44:29,047 (deepspeed_trainer:228) INFO: 16epoch:train:13301-13400batch: iter_time=1.083e-04, loss_ctc=72.862, loss_att=55.596, acc=0.700, loss=60.742, grad_norm=4.395, loss_scale=1.000, learning_rate=1.003e-04, step_time=0.366 [cnode7-012:0/16] 2024-12-06 00:45:06,000 (deepspeed_trainer:228) INFO: 16epoch:train:13401-13500batch: iter_time=1.099e-04, loss_ctc=85.523, loss_att=66.355, acc=0.677, loss=72.142, grad_norm=4.640, loss_scale=1.000, learning_rate=1.003e-04, step_time=0.369 [cnode7-012:0/16] 2024-12-06 00:45:42,591 (deepspeed_trainer:228) INFO: 16epoch:train:13501-13600batch: iter_time=1.103e-04, loss_ctc=70.648, loss_att=55.530, acc=0.695, loss=60.096, grad_norm=4.068, loss_scale=1.000, learning_rate=1.003e-04, step_time=0.365 [cnode7-012:0/16] 2024-12-06 00:46:18,621 (deepspeed_trainer:228) INFO: 16epoch:train:13601-13700batch: iter_time=1.085e-04, loss_ctc=80.326, loss_att=59.117, acc=0.691, loss=65.527, grad_norm=4.438, loss_scale=1.000, learning_rate=1.003e-04, step_time=0.360 [cnode7-012:0/16] 2024-12-06 00:46:54,471 (deepspeed_trainer:228) INFO: 16epoch:train:13701-13800batch: iter_time=1.121e-04, loss_ctc=70.699, loss_att=53.626, acc=0.688, loss=58.755, grad_norm=4.310, loss_scale=1.000, learning_rate=1.003e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-06 00:47:30,116 (deepspeed_trainer:228) INFO: 16epoch:train:13801-13900batch: iter_time=1.099e-04, loss_ctc=79.884, loss_att=66.388, acc=0.684, loss=70.413, grad_norm=4.625, loss_scale=1.000, learning_rate=1.002e-04, step_time=0.356 [2024-12-06 00:48:05,687] [INFO] [logging.py:129:log_dist] [Rank 0] step=239000, skipped=0, lr=[np.float64(0.00010020877700226041)], mom=[[0.9, 0.98]] [2024-12-06 00:48:05,688] [INFO] [timer.py:264:stop] epoch=0/micro_step=59000/global_step=59000, RunningAvgSamplesPerSec=45.01453927174509, CurrSamplesPerSec=47.166287958511504, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-06 00:48:05,690 (deepspeed_trainer:228) INFO: 16epoch:train:13901-14000batch: iter_time=1.100e-04, loss_ctc=73.218, loss_att=57.248, acc=0.686, loss=62.034, grad_norm=4.106, loss_scale=1.000, learning_rate=1.002e-04, step_time=0.356 [cnode7-012:0/16] 2024-12-06 00:48:41,276 (deepspeed_trainer:228) INFO: 16epoch:train:14001-14100batch: iter_time=1.090e-04, loss_ctc=71.113, loss_att=55.485, acc=0.693, loss=60.168, grad_norm=3.983, loss_scale=1.000, learning_rate=1.002e-04, step_time=0.355 [cnode7-012:0/16] 2024-12-06 00:49:17,232 (deepspeed_trainer:228) INFO: 16epoch:train:14101-14200batch: iter_time=1.102e-04, loss_ctc=79.876, loss_att=61.114, acc=0.706, loss=66.746, grad_norm=4.184, loss_scale=1.000, learning_rate=1.002e-04, step_time=0.359 [cnode7-012:0/16] 2024-12-06 00:49:52,903 (deepspeed_trainer:228) INFO: 16epoch:train:14201-14300batch: iter_time=1.111e-04, loss_ctc=72.933, loss_att=55.951, acc=0.690, loss=61.045, grad_norm=4.271, loss_scale=1.000, learning_rate=1.002e-04, step_time=0.356 [cnode7-012:0/16] 2024-12-06 00:50:28,589 (deepspeed_trainer:228) INFO: 16epoch:train:14301-14400batch: iter_time=1.100e-04, loss_ctc=77.241, loss_att=57.169, acc=0.682, loss=63.208, grad_norm=4.388, loss_scale=1.000, learning_rate=1.001e-04, step_time=0.356 [cnode7-012:0/16] 2024-12-06 00:51:04,351 (deepspeed_trainer:228) INFO: 16epoch:train:14401-14500batch: iter_time=1.109e-04, loss_ctc=71.864, loss_att=61.608, acc=0.672, loss=64.687, grad_norm=4.207, loss_scale=1.000, learning_rate=1.001e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-06 00:51:40,131 (deepspeed_trainer:228) INFO: 16epoch:train:14501-14600batch: iter_time=1.141e-04, loss_ctc=68.916, loss_att=58.578, acc=0.688, loss=61.684, grad_norm=4.237, loss_scale=1.000, learning_rate=1.001e-04, step_time=0.357 [cnode7-012:0/16] 2024-12-06 00:52:16,001 (deepspeed_trainer:228) INFO: 16epoch:train:14601-14700batch: iter_time=1.180e-04, loss_ctc=77.539, loss_att=56.139, acc=0.695, loss=62.569, grad_norm=4.264, loss_scale=1.000, learning_rate=1.001e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-06 00:52:51,838 (deepspeed_trainer:228) INFO: 16epoch:train:14701-14800batch: iter_time=1.231e-04, loss_ctc=72.247, loss_att=56.579, acc=0.694, loss=61.283, grad_norm=3.810, loss_scale=1.000, learning_rate=1.001e-04, step_time=0.358 [cnode7-012:0/16] 2024-12-06 00:53:27,814 (deepspeed_trainer:228) INFO: 16epoch:train:14801-14900batch: iter_time=1.230e-04, loss_ctc=77.792, loss_att=61.175, acc=0.702, loss=66.177, grad_norm=3.874, loss_scale=1.000, learning_rate=1.000e-04, step_time=0.360 [2024-12-06 00:54:03,762] [INFO] [logging.py:129:log_dist] [Rank 0] step=240000, skipped=0, lr=[np.float64(9.999979166731772e-05)], mom=[[0.9, 0.98]] [2024-12-06 00:54:03,762] [INFO] [timer.py:264:stop] epoch=0/micro_step=60000/global_step=60000, RunningAvgSamplesPerSec=45.02983202950305, CurrSamplesPerSec=48.6572829803151, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-06 00:54:03,764 (deepspeed_trainer:228) INFO: 16epoch:train:14901-15000batch: iter_time=1.166e-04, loss_ctc=81.976, loss_att=60.867, acc=0.705, loss=67.184, grad_norm=4.504, loss_scale=1.000, learning_rate=1.000e-04, step_time=0.359 [2024-12-06 00:54:20,471] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:19,170] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:20,521] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:19,358] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:20,761] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:19,443] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:20,785] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:20,859] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:20,872] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:19,605] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:20,946] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:20,978] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:19,653] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:19,770] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:19,774] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:19,800] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:36,566] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:35,913] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:35,969] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:36,095] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:37,464] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:37,581] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:37,862] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:38,250] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:37,049] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:37,100] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:38,509] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:38,530] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:37,247] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:38,590] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:37,339] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:37,401] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:52,098] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:51,709] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:53,284] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:52,066] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:52,312] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:53,678] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:54,101] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:52,920] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:52,990] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:54,440] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:53,634] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:53,839] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:53,878] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:55,328] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:55,354] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:54:55,788] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:55:07,192] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:55:08,477] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:55:07,188] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:55:08,911] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:55:07,804] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:55:08,182] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:55:08,442] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:55:09,780] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:55:08,986] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:55:10,721] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:55:09,700] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:55:09,747] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:55:11,172] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:55:10,225] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:55:11,622] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:55:11,885] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:55:24,113] [INFO] [logging.py:129:log_dist] [Rank 0] [Torch] Checkpoint 16 is about to be saved! [2024-12-06 00:55:24,143] [INFO] [logging.py:129:log_dist] [Rank 0] Saving model checkpoint: exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/mp_rank_00_model_states.pt [2024-12-06 00:55:24,144] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/mp_rank_00_model_states.pt... [2024-12-06 00:55:26,067] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/mp_rank_00_model_states.pt. [2024-12-06 00:55:26,221] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt... [2024-12-06 00:55:24,896] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_13_mp_rank_00_optim_states.pt... [2024-12-06 00:55:26,225] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt... [2024-12-06 00:55:24,897] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_11_mp_rank_00_optim_states.pt... [2024-12-06 00:55:26,225] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt... [2024-12-06 00:55:26,226] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt... [2024-12-06 00:55:26,226] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt... [2024-12-06 00:55:24,898] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_14_mp_rank_00_optim_states.pt... [2024-12-06 00:55:26,226] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt... [2024-12-06 00:55:24,899] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_10_mp_rank_00_optim_states.pt... [2024-12-06 00:55:24,899] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_15_mp_rank_00_optim_states.pt... [2024-12-06 00:55:26,227] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt... [2024-12-06 00:55:26,227] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt... [2024-12-06 00:55:24,900] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_8_mp_rank_00_optim_states.pt... [2024-12-06 00:55:24,901] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_12_mp_rank_00_optim_states.pt... [2024-12-06 00:55:24,901] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_9_mp_rank_00_optim_states.pt... [2024-12-06 00:55:26,857] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt. [2024-12-06 00:55:26,857] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt [2024-12-06 00:55:26,858] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 16 is ready now! [2024-12-06 00:55:26,989] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt. [2024-12-06 00:55:26,989] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt [2024-12-06 00:55:26,989] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 16 is ready now! [2024-12-06 00:55:26,993] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt. [2024-12-06 00:55:27,009] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt [2024-12-06 00:55:27,009] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 16 is ready now! [2024-12-06 00:55:27,012] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt. [2024-12-06 00:55:27,013] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt [2024-12-06 00:55:27,013] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 16 is ready now! [2024-12-06 00:55:27,017] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt. [2024-12-06 00:55:27,017] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt [2024-12-06 00:55:27,018] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 16 is ready now! [2024-12-06 00:55:25,702] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_12_mp_rank_00_optim_states.pt. [2024-12-06 00:55:25,702] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_12_mp_rank_00_optim_states.pt [2024-12-06 00:55:25,702] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 16 is ready now! [2024-12-06 00:55:25,704] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_14_mp_rank_00_optim_states.pt. [2024-12-06 00:55:25,705] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_14_mp_rank_00_optim_states.pt [2024-12-06 00:55:25,705] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 16 is ready now! [2024-12-06 00:55:25,707] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_10_mp_rank_00_optim_states.pt. [2024-12-06 00:55:25,707] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_10_mp_rank_00_optim_states.pt [2024-12-06 00:55:25,707] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 16 is ready now! [2024-12-06 00:55:25,724] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_8_mp_rank_00_optim_states.pt. [2024-12-06 00:55:25,724] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_8_mp_rank_00_optim_states.pt [2024-12-06 00:55:25,724] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 16 is ready now! [2024-12-06 00:55:27,061] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt. [2024-12-06 00:55:27,062] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt [2024-12-06 00:55:27,062] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 16 is ready now! [2024-12-06 00:55:25,762] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_15_mp_rank_00_optim_states.pt. [2024-12-06 00:55:25,762] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_15_mp_rank_00_optim_states.pt [2024-12-06 00:55:25,762] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 16 is ready now! [2024-12-06 00:55:27,095] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt. [2024-12-06 00:55:27,095] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt [2024-12-06 00:55:27,095] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 16 is ready now! [2024-12-06 00:55:25,772] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_9_mp_rank_00_optim_states.pt. [2024-12-06 00:55:25,772] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_9_mp_rank_00_optim_states.pt [2024-12-06 00:55:25,772] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 16 is ready now! [2024-12-06 00:55:27,117] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt. [2024-12-06 00:55:27,117] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt [2024-12-06 00:55:27,117] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 16 is ready now! [2024-12-06 00:55:25,803] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_11_mp_rank_00_optim_states.pt. [2024-12-06 00:55:25,804] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_11_mp_rank_00_optim_states.pt [2024-12-06 00:55:25,804] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 16 is ready now! [2024-12-06 00:55:25,809] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_13_mp_rank_00_optim_states.pt. [2024-12-06 00:55:25,809] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_16/16/bf16_zero_pp_rank_13_mp_rank_00_optim_states.pt [2024-12-06 00:55:25,809] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 16 is ready now! [cnode7-012:0/16] 2024-12-06 00:55:27,143 (deepspeed_trainer:158) INFO: 16epoch results: [train] iter_time=0.131, loss_ctc=76.819, loss_att=60.008, acc=0.689, loss=65.052, grad_norm=4.337, loss_scale=1.000, learning_rate=1.016e-04, step_time=0.366, time=2 hours, 4 minutes and 25 seconds, total_count=240016, gpu_max_cached_mem_GB=36.545, [valid] loss_ctc=5.094, cer_ctc=0.125, loss_att=7.500, acc=0.870, cer=0.100, wer=0.938, loss=6.781, time=1 minute and 10.91 seconds, total_count=16, gpu_max_cached_mem_GB=36.545 [cnode7-012:0/16] 2024-12-06 00:55:28,480 (multiple_iter_factory:32) INFO: Building 0th iter-factory... [cnode7-012:0/16] 2024-12-06 00:55:55,632 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-06 00:56:11,179 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.0", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.0", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.0", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.0", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-06 00:56:11,179 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.0, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-06 00:56:11,182 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-06 00:56:31,721] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:56:30,743] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:56:31,165] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:56:31,506] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:56:31,820] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:56:32,212] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:56:33,780] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:56:32,469] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:56:33,234] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:56:34,993] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:56:35,296] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:56:35,524] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:56:35,756] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:56:36,031] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:56:35,149] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:56:36,626] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:57:17,858] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:57:18,348] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:57:19,474] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:57:19,728] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:57:20,759] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:57:21,764] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:57:23,216] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:57:22,230] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:57:23,883] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:57:23,797] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:57:25,598] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:57:25,955] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:57:26,038] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:57:26,707] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:57:27,818] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:57:28,028] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:58:05,889] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:58:06,207] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:58:06,572] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:58:07,135] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:58:09,190] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:58:09,590] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:58:10,420] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:58:13,445] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:58:14,362] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:58:15,078] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:58:15,766] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:58:15,807] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:58:15,282] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:58:17,249] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:58:17,972] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:58:19,284] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:58:53,918] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:58:54,267] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:58:55,055] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:58:55,827] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:58:56,695] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:58:58,297] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:58:58,485] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:59:01,177] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:59:03,092] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:59:03,728] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:59:03,773] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:59:04,026] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:59:05,393] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:59:06,701] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:59:05,792] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 00:59:09,929] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-06 01:00:13,317 (deepspeed_trainer:228) INFO: 17epoch:train:1-100batch: iter_time=2.391, loss_ctc=83.543, loss_att=76.397, acc=0.663, loss=78.531, grad_norm=4.884, loss_scale=1.000, learning_rate=9.999e-05, step_time=0.458 [cnode7-012:0/16] 2024-12-06 01:00:49,071 (deepspeed_trainer:228) INFO: 17epoch:train:101-200batch: iter_time=1.188e-04, loss_ctc=76.701, loss_att=61.553, acc=0.676, loss=66.080, grad_norm=4.651, loss_scale=1.000, learning_rate=9.997e-05, step_time=0.357 [cnode7-012:0/16] 2024-12-06 01:01:24,840 (deepspeed_trainer:228) INFO: 17epoch:train:201-300batch: iter_time=1.109e-04, loss_ctc=77.374, loss_att=59.547, acc=0.677, loss=64.903, grad_norm=4.541, loss_scale=1.000, learning_rate=9.995e-05, step_time=0.357 [cnode7-012:0/16] 2024-12-06 01:02:00,694 (deepspeed_trainer:228) INFO: 17epoch:train:301-400batch: iter_time=1.221e-04, loss_ctc=78.854, loss_att=62.645, acc=0.699, loss=67.486, grad_norm=4.311, loss_scale=1.000, learning_rate=9.993e-05, step_time=0.358 [cnode7-012:0/16] 2024-12-06 01:02:36,354 (deepspeed_trainer:228) INFO: 17epoch:train:401-500batch: iter_time=1.196e-04, loss_ctc=73.344, loss_att=57.570, acc=0.698, loss=62.316, grad_norm=4.217, loss_scale=1.000, learning_rate=9.991e-05, step_time=0.356 [cnode7-012:0/16] 2024-12-06 01:03:12,243 (deepspeed_trainer:228) INFO: 17epoch:train:501-600batch: iter_time=1.172e-04, loss_ctc=92.235, loss_att=63.430, acc=0.690, loss=72.087, grad_norm=4.906, loss_scale=1.000, learning_rate=9.989e-05, step_time=0.359 [cnode7-012:0/16] 2024-12-06 01:03:48,332 (deepspeed_trainer:228) INFO: 17epoch:train:601-700batch: iter_time=1.134e-04, loss_ctc=78.494, loss_att=56.747, acc=0.691, loss=63.276, grad_norm=4.706, loss_scale=1.000, learning_rate=9.986e-05, step_time=0.360 [cnode7-012:0/16] 2024-12-06 01:04:24,591 (deepspeed_trainer:228) INFO: 17epoch:train:701-800batch: iter_time=1.235e-04, loss_ctc=81.938, loss_att=65.979, acc=0.687, loss=70.767, grad_norm=4.326, loss_scale=1.000, learning_rate=9.984e-05, step_time=0.362 [cnode7-012:0/16] 2024-12-06 01:05:00,544 (deepspeed_trainer:228) INFO: 17epoch:train:801-900batch: iter_time=1.198e-04, loss_ctc=77.810, loss_att=59.862, acc=0.691, loss=65.254, grad_norm=4.163, loss_scale=1.000, learning_rate=9.982e-05, step_time=0.359 [2024-12-06 01:05:37,162] [INFO] [logging.py:129:log_dist] [Rank 0] step=241000, skipped=0, lr=[np.float64(9.979210841862826e-05)], mom=[[0.9, 0.98]] [2024-12-06 01:05:37,162] [INFO] [timer.py:264:stop] epoch=0/micro_step=61000/global_step=61000, RunningAvgSamplesPerSec=45.02098048870109, CurrSamplesPerSec=46.213891137232395, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-06 01:05:37,165 (deepspeed_trainer:228) INFO: 17epoch:train:901-1000batch: iter_time=1.126e-04, loss_ctc=78.361, loss_att=62.586, acc=0.693, loss=67.304, grad_norm=4.357, loss_scale=1.000, learning_rate=9.980e-05, step_time=0.366 [cnode7-012:0/16] 2024-12-06 01:06:13,349 (deepspeed_trainer:228) INFO: 17epoch:train:1001-1100batch: iter_time=1.096e-04, loss_ctc=86.234, loss_att=72.238, acc=0.668, loss=76.438, grad_norm=4.893, loss_scale=1.000, learning_rate=9.978e-05, step_time=0.362 [cnode7-012:0/16] 2024-12-06 01:06:49,614 (deepspeed_trainer:228) INFO: 17epoch:train:1101-1200batch: iter_time=1.096e-04, loss_ctc=76.504, loss_att=65.019, acc=0.678, loss=68.470, grad_norm=4.217, loss_scale=1.000, learning_rate=9.976e-05, step_time=0.362 [cnode7-012:0/16] 2024-12-06 01:07:26,177 (deepspeed_trainer:228) INFO: 17epoch:train:1201-1300batch: iter_time=1.112e-04, loss_ctc=75.995, loss_att=61.160, acc=0.685, loss=65.577, grad_norm=4.418, loss_scale=1.000, learning_rate=9.974e-05, step_time=0.366 [cnode7-012:0/16] 2024-12-06 01:08:03,044 (deepspeed_trainer:228) INFO: 17epoch:train:1301-1400batch: iter_time=1.118e-04, loss_ctc=75.104, loss_att=58.796, acc=0.698, loss=63.698, grad_norm=4.244, loss_scale=1.000, learning_rate=9.972e-05, step_time=0.366 [cnode7-012:0/16] 2024-12-06 01:08:39,044 (deepspeed_trainer:228) INFO: 17epoch:train:1401-1500batch: iter_time=1.133e-04, loss_ctc=84.320, loss_att=65.233, acc=0.676, loss=70.994, grad_norm=4.633, loss_scale=1.000, learning_rate=9.970e-05, step_time=0.362 [cnode7-012:0/16] 2024-12-06 01:09:14,708 (deepspeed_trainer:228) INFO: 17epoch:train:1501-1600batch: iter_time=1.194e-04, loss_ctc=66.141, loss_att=48.998, acc=0.700, loss=54.148, grad_norm=4.110, loss_scale=1.000, learning_rate=9.968e-05, step_time=0.356 [cnode7-012:0/16] 2024-12-06 01:09:50,442 (deepspeed_trainer:228) INFO: 17epoch:train:1601-1700batch: iter_time=1.123e-04, loss_ctc=82.552, loss_att=61.715, acc=0.693, loss=67.972, grad_norm=4.323, loss_scale=1.000, learning_rate=9.966e-05, step_time=0.357 [cnode7-012:0/16] 2024-12-06 01:10:26,224 (deepspeed_trainer:228) INFO: 17epoch:train:1701-1800batch: iter_time=1.166e-04, loss_ctc=75.631, loss_att=61.245, acc=0.687, loss=65.582, grad_norm=4.267, loss_scale=1.000, learning_rate=9.964e-05, step_time=0.357 [cnode7-012:0/16] 2024-12-06 01:10:56,673 (multiple_iter_factory:32) INFO: Building 1th iter-factory... [cnode7-012:0/16] 2024-12-06 01:11:22,563 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-06 01:11:39,656 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.2", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.2", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.2", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.2", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-06 01:11:39,657 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.2, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-06 01:11:39,659 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-06 01:12:07,713] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:12:07,923] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:12:07,992] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:12:08,293] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:12:08,430] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:12:08,498] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:12:08,521] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:12:08,568] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:12:10,785] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:12:10,962] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:12:11,107] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:12:11,108] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:12:11,109] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:12:11,434] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:12:11,568] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:12:11,906] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:12:56,531] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:12:57,087] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:12:57,389] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:12:58,093] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:12:58,209] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:12:58,234] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:12:58,566] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:13:00,998] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:13:02,171] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:13:00,952] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:13:03,009] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:13:03,484] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:13:04,174] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:13:04,287] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:13:04,468] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:13:05,424] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:13:44,805] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:13:45,104] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:13:45,998] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:13:46,652] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:13:47,280] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:13:47,385] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:13:47,691] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:13:51,517] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:13:52,093] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:13:52,382] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:13:53,146] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:13:52,064] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:13:53,847] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:13:54,425] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:13:55,696] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:13:56,498] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:14:32,724] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:14:33,822] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:14:34,968] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:14:35,330] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:14:35,738] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:14:35,778] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:14:36,148] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:14:41,563] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:14:42,033] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:14:42,338] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:14:43,869] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:14:42,584] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:14:44,359] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:14:44,471] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:14:46,087] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:14:48,614] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-06 01:15:25,099 (deepspeed_trainer:228) INFO: 17epoch:train:1801-1900batch: iter_time=2.573, loss_ctc=75.400, loss_att=59.244, acc=0.682, loss=64.081, grad_norm=4.489, loss_scale=1.000, learning_rate=9.962e-05, step_time=0.415 [2024-12-06 01:16:01,887] [INFO] [logging.py:129:log_dist] [Rank 0] step=242000, skipped=0, lr=[np.float64(9.95857137909994e-05)], mom=[[0.9, 0.98]] [2024-12-06 01:16:01,888] [INFO] [timer.py:264:stop] epoch=0/micro_step=62000/global_step=62000, RunningAvgSamplesPerSec=45.0169882788019, CurrSamplesPerSec=43.57836741816914, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-06 01:16:01,889 (deepspeed_trainer:228) INFO: 17epoch:train:1901-2000batch: iter_time=1.114e-04, loss_ctc=78.477, loss_att=68.684, acc=0.677, loss=71.611, grad_norm=4.906, loss_scale=1.000, learning_rate=9.960e-05, step_time=0.368 [cnode7-012:0/16] 2024-12-06 01:16:38,590 (deepspeed_trainer:228) INFO: 17epoch:train:2001-2100batch: iter_time=1.185e-04, loss_ctc=79.307, loss_att=59.502, acc=0.678, loss=65.431, grad_norm=5.122, loss_scale=1.000, learning_rate=9.958e-05, step_time=0.367 [cnode7-012:0/16] 2024-12-06 01:17:15,450 (deepspeed_trainer:228) INFO: 17epoch:train:2101-2200batch: iter_time=1.151e-04, loss_ctc=74.331, loss_att=57.314, acc=0.695, loss=62.438, grad_norm=3.864, loss_scale=1.000, learning_rate=9.955e-05, step_time=0.368 [cnode7-012:0/16] 2024-12-06 01:17:52,172 (deepspeed_trainer:228) INFO: 17epoch:train:2201-2300batch: iter_time=1.122e-04, loss_ctc=79.429, loss_att=62.224, acc=0.703, loss=67.390, grad_norm=4.095, loss_scale=1.000, learning_rate=9.953e-05, step_time=0.367 [cnode7-012:0/16] 2024-12-06 01:18:28,807 (deepspeed_trainer:228) INFO: 17epoch:train:2301-2400batch: iter_time=1.097e-04, loss_ctc=73.998, loss_att=59.960, acc=0.690, loss=64.159, grad_norm=4.088, loss_scale=1.000, learning_rate=9.951e-05, step_time=0.366 [cnode7-012:0/16] 2024-12-06 01:19:05,469 (deepspeed_trainer:228) INFO: 17epoch:train:2401-2500batch: iter_time=1.098e-04, loss_ctc=88.429, loss_att=58.174, acc=0.704, loss=67.247, grad_norm=4.902, loss_scale=1.000, learning_rate=9.949e-05, step_time=0.366 [cnode7-012:0/16] 2024-12-06 01:19:42,251 (deepspeed_trainer:228) INFO: 17epoch:train:2501-2600batch: iter_time=1.129e-04, loss_ctc=77.641, loss_att=58.120, acc=0.694, loss=64.004, grad_norm=4.719, loss_scale=1.000, learning_rate=9.947e-05, step_time=0.367 [cnode7-012:0/16] 2024-12-06 01:20:18,661 (deepspeed_trainer:228) INFO: 17epoch:train:2601-2700batch: iter_time=1.177e-04, loss_ctc=77.357, loss_att=63.508, acc=0.682, loss=67.625, grad_norm=4.242, loss_scale=1.000, learning_rate=9.945e-05, step_time=0.364 [cnode7-012:0/16] 2024-12-06 01:20:55,512 (deepspeed_trainer:228) INFO: 17epoch:train:2701-2800batch: iter_time=1.152e-04, loss_ctc=79.134, loss_att=58.572, acc=0.702, loss=64.723, grad_norm=4.818, loss_scale=1.000, learning_rate=9.943e-05, step_time=0.368 [cnode7-012:0/16] 2024-12-06 01:21:32,598 (deepspeed_trainer:228) INFO: 17epoch:train:2801-2900batch: iter_time=1.131e-04, loss_ctc=81.854, loss_att=68.079, acc=0.682, loss=72.217, grad_norm=4.866, loss_scale=1.000, learning_rate=9.941e-05, step_time=0.371 [2024-12-06 01:22:09,143] [INFO] [logging.py:129:log_dist] [Rank 0] step=243000, skipped=0, lr=[np.float64(9.938059451338106e-05)], mom=[[0.9, 0.98]] [2024-12-06 01:22:09,144] [INFO] [timer.py:264:stop] epoch=0/micro_step=63000/global_step=63000, RunningAvgSamplesPerSec=45.013607501453784, CurrSamplesPerSec=45.37067560549099, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-06 01:22:09,146 (deepspeed_trainer:228) INFO: 17epoch:train:2901-3000batch: iter_time=1.204e-04, loss_ctc=80.076, loss_att=65.873, acc=0.681, loss=70.103, grad_norm=4.718, loss_scale=1.000, learning_rate=9.939e-05, step_time=0.365 [cnode7-012:0/16] 2024-12-06 01:22:45,424 (deepspeed_trainer:228) INFO: 17epoch:train:3001-3100batch: iter_time=1.127e-04, loss_ctc=76.434, loss_att=64.248, acc=0.679, loss=67.925, grad_norm=4.451, loss_scale=1.000, learning_rate=9.937e-05, step_time=0.363 [cnode7-012:0/16] 2024-12-06 01:23:21,623 (deepspeed_trainer:228) INFO: 17epoch:train:3101-3200batch: iter_time=1.151e-04, loss_ctc=73.024, loss_att=59.038, acc=0.693, loss=63.231, grad_norm=4.292, loss_scale=1.000, learning_rate=9.935e-05, step_time=0.362 [cnode7-012:0/16] 2024-12-06 01:23:57,712 (deepspeed_trainer:228) INFO: 17epoch:train:3201-3300batch: iter_time=1.149e-04, loss_ctc=78.476, loss_att=60.910, acc=0.692, loss=66.179, grad_norm=5.031, loss_scale=1.000, learning_rate=9.933e-05, step_time=0.360 [cnode7-012:0/16] 2024-12-06 01:24:34,154 (deepspeed_trainer:228) INFO: 17epoch:train:3301-3400batch: iter_time=1.120e-04, loss_ctc=78.854, loss_att=60.152, acc=0.686, loss=65.767, grad_norm=4.483, loss_scale=1.000, learning_rate=9.931e-05, step_time=0.364 [cnode7-012:0/16] 2024-12-06 01:25:10,351 (deepspeed_trainer:228) INFO: 17epoch:train:3401-3500batch: iter_time=1.144e-04, loss_ctc=67.865, loss_att=48.730, acc=0.703, loss=54.492, grad_norm=4.209, loss_scale=1.000, learning_rate=9.929e-05, step_time=0.361 [cnode7-012:0/16] 2024-12-06 01:25:46,725 (deepspeed_trainer:228) INFO: 17epoch:train:3501-3600batch: iter_time=1.141e-04, loss_ctc=81.505, loss_att=60.737, acc=0.692, loss=66.960, grad_norm=4.829, loss_scale=1.000, learning_rate=9.927e-05, step_time=0.363 [cnode7-012:0/16] 2024-12-06 01:26:23,115 (deepspeed_trainer:228) INFO: 17epoch:train:3601-3700batch: iter_time=1.186e-04, loss_ctc=74.418, loss_att=59.819, acc=0.693, loss=64.232, grad_norm=4.082, loss_scale=1.000, learning_rate=9.925e-05, step_time=0.364 [cnode7-012:0/16] 2024-12-06 01:26:43,692 (multiple_iter_factory:32) INFO: Building 2th iter-factory... [cnode7-012:0/16] 2024-12-06 01:27:10,689 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-06 01:27:27,582 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.7", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.7", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.7", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.7", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-06 01:27:27,582 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.7, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-06 01:27:27,584 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-06 01:27:53,850] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:27:54,969] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:27:55,143] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:27:54,098] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:27:54,149] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:27:54,846] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:27:55,103] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:27:55,239] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:27:55,708] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:27:55,786] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:27:55,809] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:27:58,997] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:27:59,294] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:27:59,966] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:28:00,008] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:28:00,363] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:28:41,722] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:28:43,568] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:28:42,736] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:28:42,851] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:28:43,071] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:28:43,305] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:28:43,721] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:28:44,304] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:28:44,364] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:28:47,268] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:28:48,153] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:28:48,482] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:28:47,126] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:28:49,072] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:28:49,748] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:28:49,751] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:29:30,855] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:29:31,746] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:29:31,204] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:29:31,435] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:29:31,488] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:29:32,107] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:29:32,399] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:29:35,276] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:29:36,260] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:29:37,978] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:29:37,084] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:29:38,681] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:29:38,955] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:29:39,601] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:29:39,876] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:29:42,919] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:30:18,838] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:30:18,438] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:30:19,901] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:30:18,780] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:30:20,075] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:30:20,365] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:30:20,621] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:30:25,065] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:30:27,159] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:30:26,325] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:30:28,102] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:30:28,375] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:30:27,096] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:30:30,515] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:30:32,412] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:30:34,097] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-06 01:31:18,617 (deepspeed_trainer:228) INFO: 17epoch:train:3701-3800batch: iter_time=2.550, loss_ctc=78.330, loss_att=67.132, acc=0.670, loss=70.500, grad_norm=5.146, loss_scale=1.000, learning_rate=9.923e-05, step_time=0.404 [cnode7-012:0/16] 2024-12-06 01:31:54,988 (deepspeed_trainer:228) INFO: 17epoch:train:3801-3900batch: iter_time=1.125e-04, loss_ctc=74.163, loss_att=65.438, acc=0.673, loss=68.091, grad_norm=4.687, loss_scale=1.000, learning_rate=9.921e-05, step_time=0.363 [2024-12-06 01:32:30,970] [INFO] [logging.py:129:log_dist] [Rank 0] step=244000, skipped=0, lr=[np.float64(9.91767375052817e-05)], mom=[[0.9, 0.98]] [2024-12-06 01:32:30,971] [INFO] [timer.py:264:stop] epoch=0/micro_step=64000/global_step=64000, RunningAvgSamplesPerSec=45.01156647227105, CurrSamplesPerSec=45.99827070855545, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-06 01:32:30,973 (deepspeed_trainer:228) INFO: 17epoch:train:3901-4000batch: iter_time=1.142e-04, loss_ctc=75.131, loss_att=57.408, acc=0.672, loss=62.737, grad_norm=4.690, loss_scale=1.000, learning_rate=9.919e-05, step_time=0.360 [cnode7-012:0/16] 2024-12-06 01:33:07,177 (deepspeed_trainer:228) INFO: 17epoch:train:4001-4100batch: iter_time=1.137e-04, loss_ctc=77.041, loss_att=58.395, acc=0.698, loss=63.967, grad_norm=3.958, loss_scale=1.000, learning_rate=9.917e-05, step_time=0.361 [cnode7-012:0/16] 2024-12-06 01:33:44,033 (deepspeed_trainer:228) INFO: 17epoch:train:4101-4200batch: iter_time=1.077e-04, loss_ctc=77.558, loss_att=61.646, acc=0.692, loss=66.421, grad_norm=4.539, loss_scale=1.000, learning_rate=9.915e-05, step_time=0.368 [cnode7-012:0/16] 2024-12-06 01:34:20,499 (deepspeed_trainer:228) INFO: 17epoch:train:4201-4300batch: iter_time=1.102e-04, loss_ctc=86.734, loss_att=60.330, acc=0.690, loss=68.247, grad_norm=4.604, loss_scale=1.000, learning_rate=9.913e-05, step_time=0.364 [cnode7-012:0/16] 2024-12-06 01:34:57,477 (deepspeed_trainer:228) INFO: 17epoch:train:4301-4400batch: iter_time=1.119e-04, loss_ctc=78.696, loss_att=54.784, acc=0.698, loss=61.962, grad_norm=4.438, loss_scale=1.000, learning_rate=9.911e-05, step_time=0.370 [cnode7-012:0/16] 2024-12-06 01:35:34,461 (deepspeed_trainer:228) INFO: 17epoch:train:4401-4500batch: iter_time=1.141e-04, loss_ctc=76.694, loss_att=58.835, acc=0.690, loss=64.154, grad_norm=4.071, loss_scale=1.000, learning_rate=9.909e-05, step_time=0.370 [cnode7-012:0/16] 2024-12-06 01:36:11,280 (deepspeed_trainer:228) INFO: 17epoch:train:4501-4600batch: iter_time=1.119e-04, loss_ctc=76.927, loss_att=60.173, acc=0.688, loss=65.205, grad_norm=4.544, loss_scale=1.000, learning_rate=9.907e-05, step_time=0.368 [cnode7-012:0/16] 2024-12-06 01:36:47,385 (deepspeed_trainer:228) INFO: 17epoch:train:4601-4700batch: iter_time=1.117e-04, loss_ctc=79.119, loss_att=62.055, acc=0.695, loss=67.183, grad_norm=4.382, loss_scale=1.000, learning_rate=9.904e-05, step_time=0.361 [cnode7-012:0/16] 2024-12-06 01:37:23,875 (deepspeed_trainer:228) INFO: 17epoch:train:4701-4800batch: iter_time=1.102e-04, loss_ctc=81.639, loss_att=67.637, acc=0.679, loss=71.822, grad_norm=4.945, loss_scale=1.000, learning_rate=9.902e-05, step_time=0.365 [cnode7-012:0/16] 2024-12-06 01:37:59,839 (deepspeed_trainer:228) INFO: 17epoch:train:4801-4900batch: iter_time=1.102e-04, loss_ctc=76.654, loss_att=62.292, acc=0.678, loss=66.599, grad_norm=4.395, loss_scale=1.000, learning_rate=9.900e-05, step_time=0.359 [2024-12-06 01:38:35,434] [INFO] [logging.py:129:log_dist] [Rank 0] step=245000, skipped=0, lr=[np.float64(9.897412987326465e-05)], mom=[[0.9, 0.98]] [2024-12-06 01:38:35,435] [INFO] [timer.py:264:stop] epoch=0/micro_step=65000/global_step=65000, RunningAvgSamplesPerSec=45.01352572470521, CurrSamplesPerSec=46.967171321535524, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-06 01:38:35,437 (deepspeed_trainer:228) INFO: 17epoch:train:4901-5000batch: iter_time=1.102e-04, loss_ctc=71.092, loss_att=60.812, acc=0.668, loss=63.913, grad_norm=4.273, loss_scale=1.000, learning_rate=9.898e-05, step_time=0.356 [cnode7-012:0/16] 2024-12-06 01:39:11,739 (deepspeed_trainer:228) INFO: 17epoch:train:5001-5100batch: iter_time=1.086e-04, loss_ctc=76.632, loss_att=61.410, acc=0.687, loss=66.016, grad_norm=4.207, loss_scale=1.000, learning_rate=9.896e-05, step_time=0.363 [cnode7-012:0/16] 2024-12-06 01:39:47,518 (deepspeed_trainer:228) INFO: 17epoch:train:5101-5200batch: iter_time=1.097e-04, loss_ctc=78.836, loss_att=59.182, acc=0.694, loss=65.068, grad_norm=4.039, loss_scale=1.000, learning_rate=9.894e-05, step_time=0.357 [cnode7-012:0/16] 2024-12-06 01:40:23,218 (deepspeed_trainer:228) INFO: 17epoch:train:5201-5300batch: iter_time=1.109e-04, loss_ctc=75.623, loss_att=57.104, acc=0.685, loss=62.680, grad_norm=4.331, loss_scale=1.000, learning_rate=9.892e-05, step_time=0.357 [cnode7-012:0/16] 2024-12-06 01:40:58,880 (deepspeed_trainer:228) INFO: 17epoch:train:5301-5400batch: iter_time=1.105e-04, loss_ctc=75.709, loss_att=54.846, acc=0.697, loss=61.070, grad_norm=4.200, loss_scale=1.000, learning_rate=9.890e-05, step_time=0.356 [cnode7-012:0/16] 2024-12-06 01:41:35,316 (deepspeed_trainer:228) INFO: 17epoch:train:5401-5500batch: iter_time=1.109e-04, loss_ctc=77.317, loss_att=61.229, acc=0.682, loss=66.049, grad_norm=4.326, loss_scale=1.000, learning_rate=9.888e-05, step_time=0.364 [cnode7-012:0/16] 2024-12-06 01:42:11,449 (deepspeed_trainer:228) INFO: 17epoch:train:5501-5600batch: iter_time=1.121e-04, loss_ctc=69.438, loss_att=52.820, acc=0.691, loss=57.826, grad_norm=4.337, loss_scale=1.000, learning_rate=9.886e-05, step_time=0.361 [cnode7-012:0/16] 2024-12-06 01:42:23,093 (multiple_iter_factory:32) INFO: Building 3th iter-factory... [cnode7-012:0/16] 2024-12-06 01:42:49,609 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-06 01:43:05,700 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.6", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.6", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.6", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.6", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-06 01:43:05,700 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.6, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-06 01:43:05,703 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-06 01:43:32,512] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:43:33,069] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:43:33,539] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:43:33,190] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:43:33,748] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:43:33,969] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:43:34,266] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:43:34,303] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:43:35,171] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:43:36,731] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:43:35,655] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:43:36,069] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:43:38,068] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:43:38,660] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:43:38,997] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:43:39,340] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:44:21,467] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:44:22,072] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:44:22,359] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:44:22,799] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:44:23,329] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:44:23,921] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:44:23,928] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:44:24,330] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:44:24,435] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:44:27,186] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:44:28,376] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:44:27,534] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:44:28,938] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:44:29,310] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:44:29,601] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:44:28,251] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:45:08,895] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:45:10,000] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:45:11,955] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:45:12,231] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:45:14,247] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:45:13,603] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:45:15,401] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:45:14,057] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:45:15,660] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:45:17,665] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:45:18,276] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:45:18,532] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:45:17,947] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:45:19,380] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:45:19,598] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:45:21,854] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:45:56,652] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:45:57,819] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:46:00,177] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:46:00,687] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:46:00,964] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:46:02,473] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:46:05,068] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:46:05,832] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:46:05,662] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:46:07,658] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:46:08,354] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:46:08,661] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:46:08,853] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:46:09,682] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:46:10,026] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:46:11,924] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-06 01:47:06,942 (deepspeed_trainer:228) INFO: 17epoch:train:5601-5700batch: iter_time=2.513, loss_ctc=79.775, loss_att=70.687, acc=0.673, loss=73.405, grad_norm=4.741, loss_scale=1.000, learning_rate=9.884e-05, step_time=0.442 [cnode7-012:0/16] 2024-12-06 01:47:43,988 (deepspeed_trainer:228) INFO: 17epoch:train:5701-5800batch: iter_time=1.093e-04, loss_ctc=76.277, loss_att=65.413, acc=0.674, loss=68.688, grad_norm=4.698, loss_scale=1.000, learning_rate=9.882e-05, step_time=0.370 [cnode7-012:0/16] 2024-12-06 01:48:21,133 (deepspeed_trainer:228) INFO: 17epoch:train:5801-5900batch: iter_time=1.093e-04, loss_ctc=72.100, loss_att=55.186, acc=0.684, loss=60.276, grad_norm=4.379, loss_scale=1.000, learning_rate=9.880e-05, step_time=0.371 [2024-12-06 01:48:57,677] [INFO] [logging.py:129:log_dist] [Rank 0] step=246000, skipped=0, lr=[np.float64(9.877275890752292e-05)], mom=[[0.9, 0.98]] [2024-12-06 01:48:57,678] [INFO] [timer.py:264:stop] epoch=0/micro_step=66000/global_step=66000, RunningAvgSamplesPerSec=45.003719753371364, CurrSamplesPerSec=44.35183839705479, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-06 01:48:57,680 (deepspeed_trainer:228) INFO: 17epoch:train:5901-6000batch: iter_time=1.107e-04, loss_ctc=78.767, loss_att=59.972, acc=0.708, loss=65.592, grad_norm=4.365, loss_scale=1.000, learning_rate=9.878e-05, step_time=0.365 [cnode7-012:0/16] 2024-12-06 01:49:34,373 (deepspeed_trainer:228) INFO: 17epoch:train:6001-6100batch: iter_time=1.098e-04, loss_ctc=73.862, loss_att=58.495, acc=0.709, loss=63.078, grad_norm=4.018, loss_scale=1.000, learning_rate=9.876e-05, step_time=0.366 [cnode7-012:0/16] 2024-12-06 01:50:12,014 (deepspeed_trainer:228) INFO: 17epoch:train:6101-6200batch: iter_time=1.110e-04, loss_ctc=87.888, loss_att=62.696, acc=0.696, loss=70.268, grad_norm=4.808, loss_scale=1.000, learning_rate=9.874e-05, step_time=0.376 [cnode7-012:0/16] 2024-12-06 01:50:48,866 (deepspeed_trainer:228) INFO: 17epoch:train:6201-6300batch: iter_time=1.102e-04, loss_ctc=76.502, loss_att=55.968, acc=0.696, loss=62.124, grad_norm=4.395, loss_scale=1.000, learning_rate=9.872e-05, step_time=0.368 [cnode7-012:0/16] 2024-12-06 01:51:25,651 (deepspeed_trainer:228) INFO: 17epoch:train:6301-6400batch: iter_time=1.116e-04, loss_ctc=77.935, loss_att=58.486, acc=0.695, loss=64.325, grad_norm=4.251, loss_scale=1.000, learning_rate=9.870e-05, step_time=0.368 [cnode7-012:0/16] 2024-12-06 01:52:02,191 (deepspeed_trainer:228) INFO: 17epoch:train:6401-6500batch: iter_time=1.117e-04, loss_ctc=76.895, loss_att=62.695, acc=0.691, loss=66.965, grad_norm=4.228, loss_scale=1.000, learning_rate=9.868e-05, step_time=0.365 [cnode7-012:0/16] 2024-12-06 01:52:39,014 (deepspeed_trainer:228) INFO: 17epoch:train:6501-6600batch: iter_time=1.099e-04, loss_ctc=78.750, loss_att=61.170, acc=0.701, loss=66.442, grad_norm=4.363, loss_scale=1.000, learning_rate=9.866e-05, step_time=0.368 [cnode7-012:0/16] 2024-12-06 01:53:15,827 (deepspeed_trainer:228) INFO: 17epoch:train:6601-6700batch: iter_time=1.101e-04, loss_ctc=79.329, loss_att=64.604, acc=0.676, loss=69.023, grad_norm=4.941, loss_scale=1.000, learning_rate=9.864e-05, step_time=0.368 [cnode7-012:0/16] 2024-12-06 01:53:52,551 (deepspeed_trainer:228) INFO: 17epoch:train:6701-6800batch: iter_time=1.101e-04, loss_ctc=75.865, loss_att=64.008, acc=0.690, loss=67.578, grad_norm=4.080, loss_scale=1.000, learning_rate=9.862e-05, step_time=0.367 [cnode7-012:0/16] 2024-12-06 01:54:28,960 (deepspeed_trainer:228) INFO: 17epoch:train:6801-6900batch: iter_time=1.096e-04, loss_ctc=69.832, loss_att=57.972, acc=0.686, loss=61.544, grad_norm=4.096, loss_scale=1.000, learning_rate=9.860e-05, step_time=0.364 [2024-12-06 01:55:05,442] [INFO] [logging.py:129:log_dist] [Rank 0] step=247000, skipped=0, lr=[np.float64(9.857261207853024e-05)], mom=[[0.9, 0.98]] [2024-12-06 01:55:05,443] [INFO] [timer.py:264:stop] epoch=0/micro_step=67000/global_step=67000, RunningAvgSamplesPerSec=45.00044757227887, CurrSamplesPerSec=46.09476068821429, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-06 01:55:05,445 (deepspeed_trainer:228) INFO: 17epoch:train:6901-7000batch: iter_time=1.109e-04, loss_ctc=72.509, loss_att=58.560, acc=0.701, loss=62.748, grad_norm=4.174, loss_scale=1.000, learning_rate=9.858e-05, step_time=0.364 [cnode7-012:0/16] 2024-12-06 01:55:42,309 (deepspeed_trainer:228) INFO: 17epoch:train:7001-7100batch: iter_time=1.101e-04, loss_ctc=83.541, loss_att=64.461, acc=0.688, loss=70.138, grad_norm=4.349, loss_scale=1.000, learning_rate=9.856e-05, step_time=0.368 [cnode7-012:0/16] 2024-12-06 01:56:18,990 (deepspeed_trainer:228) INFO: 17epoch:train:7101-7200batch: iter_time=1.111e-04, loss_ctc=71.674, loss_att=55.572, acc=0.692, loss=60.427, grad_norm=4.368, loss_scale=1.000, learning_rate=9.854e-05, step_time=0.367 [cnode7-012:0/16] 2024-12-06 01:56:55,690 (deepspeed_trainer:228) INFO: 17epoch:train:7201-7300batch: iter_time=1.119e-04, loss_ctc=77.391, loss_att=56.508, acc=0.699, loss=62.784, grad_norm=4.333, loss_scale=1.000, learning_rate=9.852e-05, step_time=0.367 [cnode7-012:0/16] 2024-12-06 01:57:32,390 (deepspeed_trainer:228) INFO: 17epoch:train:7301-7400batch: iter_time=1.098e-04, loss_ctc=75.743, loss_att=62.813, acc=0.689, loss=66.707, grad_norm=4.430, loss_scale=1.000, learning_rate=9.850e-05, step_time=0.367 [cnode7-012:0/16] 2024-12-06 01:58:08,857 (deepspeed_trainer:228) INFO: 17epoch:train:7401-7500batch: iter_time=1.093e-04, loss_ctc=71.230, loss_att=52.157, acc=0.694, loss=57.855, grad_norm=4.512, loss_scale=1.000, learning_rate=9.848e-05, step_time=0.365 [cnode7-012:0/16] 2024-12-06 01:58:11,567 (multiple_iter_factory:32) INFO: Building 4th iter-factory... [cnode7-012:0/16] 2024-12-06 01:58:38,702 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-06 01:58:55,064 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.3", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.3", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.3", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.3", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-06 01:58:55,064 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.3, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-06 01:58:55,066 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-06 01:59:21,250] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:59:21,367] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:59:21,689] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:59:20,993] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:59:21,171] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:59:21,614] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:59:21,824] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:59:22,134] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:59:23,935] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:59:22,732] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:59:22,867] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:59:22,906] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:59:25,172] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:59:26,152] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:59:26,364] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 01:59:31,354] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:00:09,062] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:00:10,299] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:00:10,934] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:00:11,215] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:00:11,470] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:00:13,033] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:00:13,288] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:00:11,912] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:00:11,947] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:00:12,076] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:00:13,528] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:00:15,119] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:00:13,772] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:00:15,521] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:00:16,551] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:00:20,198] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:00:59,039] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:00:58,417] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:00:58,752] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:00:59,240] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:00:59,643] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:01:00,324] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:01:00,858] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:01:01,190] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:01:02,638] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:01:03,583] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:01:04,862] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:01:04,925] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:01:05,182] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:01:04,087] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:01:05,766] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:01:08,856] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:01:46,034] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:01:46,523] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:01:47,756] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:01:48,276] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:01:48,556] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:01:50,178] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:01:48,918] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:01:49,185] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:01:52,358] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:01:54,152] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:01:54,964] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:01:54,095] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:01:55,514] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:01:56,413] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:01:58,424] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:02:12,477] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-06 02:03:14,654 (deepspeed_trainer:228) INFO: 17epoch:train:7501-7600batch: iter_time=2.544, loss_ctc=79.846, loss_att=73.078, acc=0.666, loss=75.141, grad_norm=4.865, loss_scale=1.000, learning_rate=9.846e-05, step_time=0.514 [cnode7-012:0/16] 2024-12-06 02:03:50,935 (deepspeed_trainer:228) INFO: 17epoch:train:7601-7700batch: iter_time=1.092e-04, loss_ctc=73.277, loss_att=58.218, acc=0.686, loss=62.719, grad_norm=4.487, loss_scale=1.000, learning_rate=9.844e-05, step_time=0.363 [cnode7-012:0/16] 2024-12-06 02:04:27,782 (deepspeed_trainer:228) INFO: 17epoch:train:7701-7800batch: iter_time=1.147e-04, loss_ctc=74.856, loss_att=57.362, acc=0.682, loss=62.604, grad_norm=4.576, loss_scale=1.000, learning_rate=9.842e-05, step_time=0.368 [cnode7-012:0/16] 2024-12-06 02:05:03,951 (deepspeed_trainer:228) INFO: 17epoch:train:7801-7900batch: iter_time=1.124e-04, loss_ctc=76.805, loss_att=58.801, acc=0.703, loss=64.207, grad_norm=4.416, loss_scale=1.000, learning_rate=9.840e-05, step_time=0.361 [2024-12-06 02:05:39,977] [INFO] [logging.py:129:log_dist] [Rank 0] step=248000, skipped=0, lr=[np.float64(9.83736770337666e-05)], mom=[[0.9, 0.98]] [2024-12-06 02:05:39,977] [INFO] [timer.py:264:stop] epoch=0/micro_step=68000/global_step=68000, RunningAvgSamplesPerSec=44.97354044689455, CurrSamplesPerSec=43.83999109056002, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-06 02:05:39,979 (deepspeed_trainer:228) INFO: 17epoch:train:7901-8000batch: iter_time=1.150e-04, loss_ctc=70.803, loss_att=56.788, acc=0.697, loss=61.008, grad_norm=4.147, loss_scale=1.000, learning_rate=9.838e-05, step_time=0.360 [cnode7-012:0/16] 2024-12-06 02:06:16,403 (deepspeed_trainer:228) INFO: 17epoch:train:8001-8100batch: iter_time=1.130e-04, loss_ctc=89.277, loss_att=58.393, acc=0.701, loss=67.648, grad_norm=4.706, loss_scale=1.000, learning_rate=9.836e-05, step_time=0.364 [cnode7-012:0/16] 2024-12-06 02:06:52,654 (deepspeed_trainer:228) INFO: 17epoch:train:8101-8200batch: iter_time=1.138e-04, loss_ctc=76.536, loss_att=55.176, acc=0.693, loss=61.601, grad_norm=4.675, loss_scale=1.000, learning_rate=9.834e-05, step_time=0.362 [cnode7-012:0/16] 2024-12-06 02:07:29,878 (deepspeed_trainer:228) INFO: 17epoch:train:8201-8300batch: iter_time=1.110e-04, loss_ctc=78.649, loss_att=62.005, acc=0.691, loss=66.999, grad_norm=4.161, loss_scale=1.000, learning_rate=9.832e-05, step_time=0.372 [cnode7-012:0/16] 2024-12-06 02:08:05,816 (deepspeed_trainer:228) INFO: 17epoch:train:8301-8400batch: iter_time=1.123e-04, loss_ctc=75.484, loss_att=58.400, acc=0.691, loss=63.505, grad_norm=4.381, loss_scale=1.000, learning_rate=9.830e-05, step_time=0.359 [cnode7-012:0/16] 2024-12-06 02:08:42,198 (deepspeed_trainer:228) INFO: 17epoch:train:8401-8500batch: iter_time=1.131e-04, loss_ctc=76.256, loss_att=61.598, acc=0.690, loss=66.015, grad_norm=4.556, loss_scale=1.000, learning_rate=9.828e-05, step_time=0.363 [cnode7-012:0/16] 2024-12-06 02:09:18,706 (deepspeed_trainer:228) INFO: 17epoch:train:8501-8600batch: iter_time=1.162e-04, loss_ctc=82.849, loss_att=69.328, acc=0.671, loss=73.370, grad_norm=5.382, loss_scale=1.000, learning_rate=9.826e-05, step_time=0.365 [cnode7-012:0/16] 2024-12-06 02:09:54,939 (deepspeed_trainer:228) INFO: 17epoch:train:8601-8700batch: iter_time=1.131e-04, loss_ctc=74.489, loss_att=62.558, acc=0.676, loss=66.151, grad_norm=4.268, loss_scale=1.000, learning_rate=9.824e-05, step_time=0.361 [cnode7-012:0/16] 2024-12-06 02:10:30,926 (deepspeed_trainer:228) INFO: 17epoch:train:8701-8800batch: iter_time=1.122e-04, loss_ctc=74.324, loss_att=58.981, acc=0.688, loss=63.596, grad_norm=4.445, loss_scale=1.000, learning_rate=9.823e-05, step_time=0.360 [cnode7-012:0/16] 2024-12-06 02:11:07,062 (deepspeed_trainer:228) INFO: 17epoch:train:8801-8900batch: iter_time=1.125e-04, loss_ctc=73.320, loss_att=57.438, acc=0.696, loss=62.204, grad_norm=4.124, loss_scale=1.000, learning_rate=9.821e-05, step_time=0.361 [2024-12-06 02:11:43,364] [INFO] [logging.py:129:log_dist] [Rank 0] step=249000, skipped=0, lr=[np.float64(9.817594159451616e-05)], mom=[[0.9, 0.98]] [2024-12-06 02:11:43,365] [INFO] [timer.py:264:stop] epoch=0/micro_step=69000/global_step=69000, RunningAvgSamplesPerSec=44.97822762728229, CurrSamplesPerSec=46.42325572917834, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-06 02:11:43,367 (deepspeed_trainer:228) INFO: 17epoch:train:8901-9000batch: iter_time=1.108e-04, loss_ctc=81.444, loss_att=62.583, acc=0.678, loss=68.212, grad_norm=4.697, loss_scale=1.000, learning_rate=9.819e-05, step_time=0.363 [cnode7-012:0/16] 2024-12-06 02:12:19,412 (deepspeed_trainer:228) INFO: 17epoch:train:9001-9100batch: iter_time=1.149e-04, loss_ctc=64.666, loss_att=47.356, acc=0.701, loss=52.572, grad_norm=4.395, loss_scale=1.000, learning_rate=9.817e-05, step_time=0.360 [cnode7-012:0/16] 2024-12-06 02:12:55,772 (deepspeed_trainer:228) INFO: 17epoch:train:9101-9200batch: iter_time=1.124e-04, loss_ctc=81.158, loss_att=61.117, acc=0.693, loss=67.120, grad_norm=4.677, loss_scale=1.000, learning_rate=9.815e-05, step_time=0.363 [cnode7-012:0/16] 2024-12-06 02:13:31,870 (deepspeed_trainer:228) INFO: 17epoch:train:9201-9300batch: iter_time=1.139e-04, loss_ctc=73.798, loss_att=57.343, acc=0.691, loss=62.226, grad_norm=4.639, loss_scale=1.000, learning_rate=9.813e-05, step_time=0.361 [cnode7-012:0/16] 2024-12-06 02:14:01,796 (multiple_iter_factory:32) INFO: Building 5th iter-factory... [cnode7-012:0/16] 2024-12-06 02:14:28,688 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-06 02:14:45,124 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.5", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.5", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.5", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.5", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-06 02:14:45,124 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.5, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-06 02:14:45,126 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-06 02:15:11,687] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:15:11,165] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:15:12,616] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:15:11,553] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:15:12,136] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:15:14,088] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:15:12,910] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:15:13,043] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:15:13,110] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:15:13,194] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:15:13,564] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:15:16,538] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:15:18,124] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:15:18,467] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:15:18,884] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:15:19,284] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:16:00,750] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:16:01,297] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:15:59,983] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:16:00,662] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:16:02,077] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:16:03,713] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:16:02,827] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:16:02,844] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:16:02,929] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:16:03,154] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:16:04,476] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:16:05,974] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:16:07,473] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:16:07,768] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:16:08,597] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:16:09,077] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:16:47,653] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:16:49,609] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:16:50,801] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:16:49,712] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:16:52,273] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:16:50,850] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:16:51,256] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:16:52,006] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:16:52,535] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:16:52,620] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:16:55,812] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:16:55,128] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:16:57,971] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:16:58,390] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:16:59,234] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:16:59,924] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:17:35,366] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:17:37,194] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:17:37,583] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:17:37,232] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:17:40,076] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:17:39,393] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:17:40,419] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:17:40,681] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:17:41,021] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:17:41,231] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:17:44,406] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:17:46,483] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:17:45,551] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:17:47,651] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:17:47,692] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:17:48,512] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-06 02:18:24,636 (deepspeed_trainer:228) INFO: 17epoch:train:9301-9400batch: iter_time=2.480, loss_ctc=73.519, loss_att=58.210, acc=0.680, loss=62.815, grad_norm=4.718, loss_scale=1.000, learning_rate=9.811e-05, step_time=0.446 [cnode7-012:0/16] 2024-12-06 02:19:01,484 (deepspeed_trainer:228) INFO: 17epoch:train:9401-9500batch: iter_time=1.103e-04, loss_ctc=76.336, loss_att=67.074, acc=0.675, loss=69.820, grad_norm=4.837, loss_scale=1.000, learning_rate=9.809e-05, step_time=0.369 [cnode7-012:0/16] 2024-12-06 02:19:37,774 (deepspeed_trainer:228) INFO: 17epoch:train:9501-9600batch: iter_time=1.118e-04, loss_ctc=76.438, loss_att=57.101, acc=0.682, loss=62.930, grad_norm=4.854, loss_scale=1.000, learning_rate=9.807e-05, step_time=0.363 [cnode7-012:0/16] 2024-12-06 02:20:14,083 (deepspeed_trainer:228) INFO: 17epoch:train:9601-9700batch: iter_time=1.149e-04, loss_ctc=73.686, loss_att=55.972, acc=0.695, loss=61.263, grad_norm=4.244, loss_scale=1.000, learning_rate=9.805e-05, step_time=0.363 [cnode7-012:0/16] 2024-12-06 02:20:50,517 (deepspeed_trainer:228) INFO: 17epoch:train:9701-9800batch: iter_time=1.176e-04, loss_ctc=78.547, loss_att=59.774, acc=0.701, loss=65.372, grad_norm=4.394, loss_scale=1.000, learning_rate=9.803e-05, step_time=0.364 [cnode7-012:0/16] 2024-12-06 02:21:26,819 (deepspeed_trainer:228) INFO: 17epoch:train:9801-9900batch: iter_time=1.111e-04, loss_ctc=73.147, loss_att=57.176, acc=0.692, loss=61.966, grad_norm=4.457, loss_scale=1.000, learning_rate=9.801e-05, step_time=0.362 [2024-12-06 02:22:03,128] [INFO] [logging.py:129:log_dist] [Rank 0] step=250000, skipped=0, lr=[np.float64(9.797939375273558e-05)], mom=[[0.9, 0.98]] [2024-12-06 02:22:03,129] [INFO] [timer.py:264:stop] epoch=0/micro_step=70000/global_step=70000, RunningAvgSamplesPerSec=44.9677426743154, CurrSamplesPerSec=44.75937858908089, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-06 02:22:03,131 (deepspeed_trainer:228) INFO: 17epoch:train:9901-10000batch: iter_time=1.112e-04, loss_ctc=86.020, loss_att=55.753, acc=0.708, loss=64.841, grad_norm=5.176, loss_scale=1.000, learning_rate=9.799e-05, step_time=0.363 [cnode7-012:0/16] 2024-12-06 02:22:39,830 (deepspeed_trainer:228) INFO: 17epoch:train:10001-10100batch: iter_time=1.114e-04, loss_ctc=76.289, loss_att=57.173, acc=0.696, loss=62.925, grad_norm=4.233, loss_scale=1.000, learning_rate=9.797e-05, step_time=0.367 [cnode7-012:0/16] 2024-12-06 02:23:16,165 (deepspeed_trainer:228) INFO: 17epoch:train:10101-10200batch: iter_time=1.124e-04, loss_ctc=76.031, loss_att=60.309, acc=0.685, loss=65.032, grad_norm=4.088, loss_scale=1.000, learning_rate=9.795e-05, step_time=0.363 [cnode7-012:0/16] 2024-12-06 02:23:52,696 (deepspeed_trainer:228) INFO: 17epoch:train:10201-10300batch: iter_time=1.119e-04, loss_ctc=78.146, loss_att=58.007, acc=0.703, loss=64.020, grad_norm=4.764, loss_scale=1.000, learning_rate=9.793e-05, step_time=0.364 [cnode7-012:0/16] 2024-12-06 02:24:29,331 (deepspeed_trainer:228) INFO: 17epoch:train:10301-10400batch: iter_time=1.121e-04, loss_ctc=79.829, loss_att=67.118, acc=0.677, loss=70.953, grad_norm=4.996, loss_scale=1.000, learning_rate=9.791e-05, step_time=0.367 [cnode7-012:0/16] 2024-12-06 02:25:05,562 (deepspeed_trainer:228) INFO: 17epoch:train:10401-10500batch: iter_time=1.126e-04, loss_ctc=78.170, loss_att=64.402, acc=0.681, loss=68.555, grad_norm=4.413, loss_scale=1.000, learning_rate=9.789e-05, step_time=0.362 [cnode7-012:0/16] 2024-12-06 02:25:42,079 (deepspeed_trainer:228) INFO: 17epoch:train:10501-10600batch: iter_time=1.110e-04, loss_ctc=75.021, loss_att=62.746, acc=0.673, loss=66.430, grad_norm=4.668, loss_scale=1.000, learning_rate=9.787e-05, step_time=0.365 [cnode7-012:0/16] 2024-12-06 02:26:19,105 (deepspeed_trainer:228) INFO: 17epoch:train:10601-10700batch: iter_time=1.121e-04, loss_ctc=71.969, loss_att=58.142, acc=0.694, loss=62.295, grad_norm=4.322, loss_scale=1.000, learning_rate=9.785e-05, step_time=0.370 [cnode7-012:0/16] 2024-12-06 02:26:55,078 (deepspeed_trainer:228) INFO: 17epoch:train:10701-10800batch: iter_time=1.115e-04, loss_ctc=76.573, loss_att=57.984, acc=0.691, loss=63.553, grad_norm=4.285, loss_scale=1.000, learning_rate=9.783e-05, step_time=0.359 [cnode7-012:0/16] 2024-12-06 02:27:30,883 (deepspeed_trainer:228) INFO: 17epoch:train:10801-10900batch: iter_time=1.095e-04, loss_ctc=77.873, loss_att=58.190, acc=0.689, loss=64.094, grad_norm=4.587, loss_scale=1.000, learning_rate=9.781e-05, step_time=0.358 [2024-12-06 02:28:06,431] [INFO] [logging.py:129:log_dist] [Rank 0] step=251000, skipped=0, lr=[np.float64(9.778402166799114e-05)], mom=[[0.9, 0.98]] [2024-12-06 02:28:06,432] [INFO] [timer.py:264:stop] epoch=0/micro_step=71000/global_step=71000, RunningAvgSamplesPerSec=44.97278127937876, CurrSamplesPerSec=45.09852171444296, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-06 02:28:06,434 (deepspeed_trainer:228) INFO: 17epoch:train:10901-11000batch: iter_time=1.106e-04, loss_ctc=67.388, loss_att=48.618, acc=0.701, loss=54.240, grad_norm=4.430, loss_scale=1.000, learning_rate=9.779e-05, step_time=0.355 [cnode7-012:0/16] 2024-12-06 02:28:42,259 (deepspeed_trainer:228) INFO: 17epoch:train:11001-11100batch: iter_time=1.119e-04, loss_ctc=79.761, loss_att=60.522, acc=0.691, loss=66.322, grad_norm=4.803, loss_scale=1.000, learning_rate=9.777e-05, step_time=0.358 [cnode7-012:0/16] 2024-12-06 02:29:18,026 (deepspeed_trainer:228) INFO: 17epoch:train:11101-11200batch: iter_time=1.109e-04, loss_ctc=73.353, loss_att=57.325, acc=0.693, loss=62.117, grad_norm=4.264, loss_scale=1.000, learning_rate=9.775e-05, step_time=0.357 [cnode7-012:0/16] 2024-12-06 02:29:38,588 (multiple_iter_factory:32) INFO: Building 6th iter-factory... [cnode7-012:0/16] 2024-12-06 02:30:06,055 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-06 02:30:22,302 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.4", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.4", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.4", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.4", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-06 02:30:22,302 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.4, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-06 02:30:22,304 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-06 02:30:47,515] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:30:49,397] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:30:48,049] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:30:49,668] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:30:48,905] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:30:49,147] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:30:49,661] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:30:49,701] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:30:51,209] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:30:49,963] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:30:51,486] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:30:51,595] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:30:50,569] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:30:52,043] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:30:50,615] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:30:52,309] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:31:36,829] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:31:35,799] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:31:37,514] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:31:37,662] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:31:37,703] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:31:38,194] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:31:38,501] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:31:38,683] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:31:38,843] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:31:40,666] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:31:40,688] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:31:39,420] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:31:41,176] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:31:41,287] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:31:41,558] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:31:41,808] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:32:24,665] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:32:23,603] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:32:25,737] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:32:24,835] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:32:25,981] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:32:27,144] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:32:27,184] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:32:27,918] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:32:28,024] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:32:28,067] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:32:30,283] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:32:30,788] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:32:30,848] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:32:31,142] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:32:31,496] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:32:33,943] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:33:12,365] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:33:11,630] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:33:14,305] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:33:13,486] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:33:14,369] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:33:14,747] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:33:15,105] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:33:15,926] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:33:16,450] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:33:16,461] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:33:18,914] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:33:20,188] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:33:20,433] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:33:20,939] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:33:21,484] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:33:25,165] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-06 02:34:09,601 (deepspeed_trainer:228) INFO: 17epoch:train:11201-11300batch: iter_time=2.552, loss_ctc=77.284, loss_att=65.586, acc=0.676, loss=69.108, grad_norm=4.784, loss_scale=1.000, learning_rate=9.774e-05, step_time=0.364 [cnode7-012:0/16] 2024-12-06 02:34:46,133 (deepspeed_trainer:228) INFO: 17epoch:train:11301-11400batch: iter_time=1.065e-04, loss_ctc=73.364, loss_att=67.118, acc=0.680, loss=68.980, grad_norm=4.511, loss_scale=1.000, learning_rate=9.772e-05, step_time=0.365 [cnode7-012:0/16] 2024-12-06 02:35:22,025 (deepspeed_trainer:228) INFO: 17epoch:train:11401-11500batch: iter_time=1.084e-04, loss_ctc=73.322, loss_att=55.325, acc=0.687, loss=60.724, grad_norm=4.657, loss_scale=1.000, learning_rate=9.770e-05, step_time=0.359 [cnode7-012:0/16] 2024-12-06 02:35:58,551 (deepspeed_trainer:228) INFO: 17epoch:train:11501-11600batch: iter_time=1.121e-04, loss_ctc=75.847, loss_att=59.087, acc=0.704, loss=64.132, grad_norm=4.339, loss_scale=1.000, learning_rate=9.768e-05, step_time=0.365 [cnode7-012:0/16] 2024-12-06 02:36:34,713 (deepspeed_trainer:228) INFO: 17epoch:train:11601-11700batch: iter_time=1.197e-04, loss_ctc=76.570, loss_att=60.808, acc=0.705, loss=65.563, grad_norm=4.345, loss_scale=1.000, learning_rate=9.766e-05, step_time=0.362 [cnode7-012:0/16] 2024-12-06 02:37:11,268 (deepspeed_trainer:228) INFO: 17epoch:train:11701-11800batch: iter_time=1.138e-04, loss_ctc=85.049, loss_att=60.353, acc=0.698, loss=67.744, grad_norm=4.500, loss_scale=1.000, learning_rate=9.764e-05, step_time=0.365 [cnode7-012:0/16] 2024-12-06 02:37:47,581 (deepspeed_trainer:228) INFO: 17epoch:train:11801-11900batch: iter_time=1.096e-04, loss_ctc=77.718, loss_att=56.108, acc=0.700, loss=62.581, grad_norm=4.584, loss_scale=1.000, learning_rate=9.762e-05, step_time=0.363 [2024-12-06 02:38:23,959] [INFO] [logging.py:129:log_dist] [Rank 0] step=252000, skipped=0, lr=[np.float64(9.758981366446274e-05)], mom=[[0.9, 0.98]] [2024-12-06 02:38:23,960] [INFO] [timer.py:264:stop] epoch=0/micro_step=72000/global_step=72000, RunningAvgSamplesPerSec=44.97866891834401, CurrSamplesPerSec=46.24420818763532, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-06 02:38:23,963 (deepspeed_trainer:228) INFO: 17epoch:train:11901-12000batch: iter_time=1.104e-04, loss_ctc=76.041, loss_att=59.081, acc=0.693, loss=64.156, grad_norm=4.382, loss_scale=1.000, learning_rate=9.760e-05, step_time=0.363 [cnode7-012:0/16] 2024-12-06 02:39:00,145 (deepspeed_trainer:228) INFO: 17epoch:train:12001-12100batch: iter_time=1.107e-04, loss_ctc=74.184, loss_att=60.667, acc=0.697, loss=64.708, grad_norm=4.306, loss_scale=1.000, learning_rate=9.758e-05, step_time=0.362 [cnode7-012:0/16] 2024-12-06 02:39:36,446 (deepspeed_trainer:228) INFO: 17epoch:train:12101-12200batch: iter_time=1.097e-04, loss_ctc=78.198, loss_att=60.586, acc=0.705, loss=65.865, grad_norm=4.394, loss_scale=1.000, learning_rate=9.756e-05, step_time=0.363 [cnode7-012:0/16] 2024-12-06 02:40:12,988 (deepspeed_trainer:228) INFO: 17epoch:train:12201-12300batch: iter_time=1.126e-04, loss_ctc=78.712, loss_att=67.065, acc=0.689, loss=70.546, grad_norm=5.045, loss_scale=1.000, learning_rate=9.754e-05, step_time=0.365 [cnode7-012:0/16] 2024-12-06 02:40:49,267 (deepspeed_trainer:228) INFO: 17epoch:train:12301-12400batch: iter_time=1.126e-04, loss_ctc=75.371, loss_att=61.664, acc=0.686, loss=65.759, grad_norm=4.494, loss_scale=1.000, learning_rate=9.752e-05, step_time=0.362 [cnode7-012:0/16] 2024-12-06 02:41:25,489 (deepspeed_trainer:228) INFO: 17epoch:train:12401-12500batch: iter_time=1.096e-04, loss_ctc=70.463, loss_att=60.552, acc=0.680, loss=63.539, grad_norm=4.527, loss_scale=1.000, learning_rate=9.750e-05, step_time=0.362 [cnode7-012:0/16] 2024-12-06 02:42:01,808 (deepspeed_trainer:228) INFO: 17epoch:train:12501-12600batch: iter_time=1.121e-04, loss_ctc=75.227, loss_att=60.450, acc=0.696, loss=64.915, grad_norm=4.357, loss_scale=1.000, learning_rate=9.748e-05, step_time=0.363 [cnode7-012:0/16] 2024-12-06 02:42:38,075 (deepspeed_trainer:228) INFO: 17epoch:train:12601-12700batch: iter_time=1.097e-04, loss_ctc=76.804, loss_att=59.737, acc=0.701, loss=64.851, grad_norm=4.413, loss_scale=1.000, learning_rate=9.746e-05, step_time=0.362 [cnode7-012:0/16] 2024-12-06 02:43:14,445 (deepspeed_trainer:228) INFO: 17epoch:train:12701-12800batch: iter_time=1.088e-04, loss_ctc=74.546, loss_att=56.536, acc=0.694, loss=61.946, grad_norm=4.714, loss_scale=1.000, learning_rate=9.744e-05, step_time=0.363 [cnode7-012:0/16] 2024-12-06 02:43:50,701 (deepspeed_trainer:228) INFO: 17epoch:train:12801-12900batch: iter_time=1.105e-04, loss_ctc=74.976, loss_att=54.092, acc=0.706, loss=60.362, grad_norm=4.331, loss_scale=1.000, learning_rate=9.743e-05, step_time=0.362 [2024-12-06 02:44:27,158] [INFO] [logging.py:129:log_dist] [Rank 0] step=253000, skipped=0, lr=[np.float64(9.739675822801309e-05)], mom=[[0.9, 0.98]] [2024-12-06 02:44:27,159] [INFO] [timer.py:264:stop] epoch=0/micro_step=73000/global_step=73000, RunningAvgSamplesPerSec=44.98294374648707, CurrSamplesPerSec=45.657748906343045, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-06 02:44:27,161 (deepspeed_trainer:228) INFO: 17epoch:train:12901-13000batch: iter_time=1.153e-04, loss_ctc=76.903, loss_att=62.622, acc=0.686, loss=66.889, grad_norm=4.379, loss_scale=1.000, learning_rate=9.741e-05, step_time=0.364 [cnode7-012:0/16] 2024-12-06 02:45:03,490 (deepspeed_trainer:228) INFO: 17epoch:train:13001-13100batch: iter_time=1.105e-04, loss_ctc=68.661, loss_att=52.218, acc=0.700, loss=57.172, grad_norm=4.172, loss_scale=1.000, learning_rate=9.739e-05, step_time=0.363 [cnode7-012:0/16] 2024-12-06 02:45:15,147 (multiple_iter_factory:32) INFO: Building 7th iter-factory... [cnode7-012:0/16] 2024-12-06 02:45:41,773 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-06 02:45:58,058 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.1", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.1", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.1", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.1", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-06 02:45:58,058 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.1, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-06 02:45:58,060 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-06 02:46:26,352] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:46:25,103] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:46:26,702] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:46:26,856] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:46:25,676] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:46:26,632] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:46:26,825] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:46:28,467] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:46:27,040] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:46:27,171] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:46:27,284] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:46:28,782] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:46:27,693] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:46:29,448] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:46:30,487] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:46:30,556] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:47:14,614] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:47:14,081] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:47:15,750] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:47:16,269] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:47:14,931] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:47:15,294] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:47:15,637] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:47:16,454] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:47:16,489] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:47:16,597] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:47:18,122] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:47:18,609] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:47:19,782] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:47:20,157] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:47:20,599] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:47:19,191] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:48:02,543] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:48:03,035] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:48:04,574] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:48:04,620] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:48:04,034] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:48:04,280] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:48:04,573] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:48:07,320] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:48:05,942] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:48:07,552] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:48:06,256] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:48:09,411] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:48:09,648] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:48:10,153] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:48:08,980] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:48:10,403] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:48:51,302] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:48:52,540] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:48:52,923] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:48:51,474] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:48:52,983] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:48:53,108] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:48:54,526] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:48:54,672] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:48:54,908] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:48:58,414] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:48:58,694] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:48:59,626] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:49:00,417] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:48:59,018] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:49:01,140] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 02:49:00,820] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-06 02:49:56,080 (deepspeed_trainer:228) INFO: 17epoch:train:13101-13200batch: iter_time=2.479, loss_ctc=77.552, loss_att=69.166, acc=0.675, loss=71.680, grad_norm=4.700, loss_scale=1.000, learning_rate=9.737e-05, step_time=0.446 [cnode7-012:0/16] 2024-12-06 02:50:33,010 (deepspeed_trainer:228) INFO: 17epoch:train:13201-13300batch: iter_time=1.172e-04, loss_ctc=74.749, loss_att=63.680, acc=0.673, loss=66.977, grad_norm=4.974, loss_scale=1.000, learning_rate=9.735e-05, step_time=0.369 [cnode7-012:0/16] 2024-12-06 02:51:10,090 (deepspeed_trainer:228) INFO: 17epoch:train:13301-13400batch: iter_time=1.350e-04, loss_ctc=71.261, loss_att=54.947, acc=0.685, loss=59.831, grad_norm=4.711, loss_scale=1.000, learning_rate=9.733e-05, step_time=0.370 [cnode7-012:0/16] 2024-12-06 02:51:46,614 (deepspeed_trainer:228) INFO: 17epoch:train:13401-13500batch: iter_time=1.161e-04, loss_ctc=77.529, loss_att=57.100, acc=0.707, loss=63.231, grad_norm=4.314, loss_scale=1.000, learning_rate=9.731e-05, step_time=0.365 [cnode7-012:0/16] 2024-12-06 02:52:23,244 (deepspeed_trainer:228) INFO: 17epoch:train:13501-13600batch: iter_time=1.101e-04, loss_ctc=74.131, loss_att=57.829, acc=0.703, loss=62.767, grad_norm=4.248, loss_scale=1.000, learning_rate=9.729e-05, step_time=0.366 [cnode7-012:0/16] 2024-12-06 02:52:59,607 (deepspeed_trainer:228) INFO: 17epoch:train:13601-13700batch: iter_time=1.089e-04, loss_ctc=86.760, loss_att=59.580, acc=0.697, loss=67.717, grad_norm=4.436, loss_scale=1.000, learning_rate=9.727e-05, step_time=0.363 [cnode7-012:0/16] 2024-12-06 02:53:35,930 (deepspeed_trainer:228) INFO: 17epoch:train:13701-13800batch: iter_time=1.218e-04, loss_ctc=75.904, loss_att=55.274, acc=0.697, loss=61.459, grad_norm=5.076, loss_scale=1.000, learning_rate=9.725e-05, step_time=0.363 [cnode7-012:0/16] 2024-12-06 02:54:12,151 (deepspeed_trainer:228) INFO: 17epoch:train:13801-13900batch: iter_time=1.100e-04, loss_ctc=76.546, loss_att=55.947, acc=0.700, loss=62.102, grad_norm=4.509, loss_scale=1.000, learning_rate=9.723e-05, step_time=0.362 [2024-12-06 02:54:48,194] [INFO] [logging.py:129:log_dist] [Rank 0] step=254000, skipped=0, lr=[np.float64(9.720484400332038e-05)], mom=[[0.9, 0.98]] [2024-12-06 02:54:48,194] [INFO] [timer.py:264:stop] epoch=0/micro_step=74000/global_step=74000, RunningAvgSamplesPerSec=44.971258851100245, CurrSamplesPerSec=47.306230281304686, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-06 02:54:48,196 (deepspeed_trainer:228) INFO: 17epoch:train:13901-14000batch: iter_time=1.119e-04, loss_ctc=76.344, loss_att=62.187, acc=0.689, loss=66.437, grad_norm=4.733, loss_scale=1.000, learning_rate=9.721e-05, step_time=0.360 [cnode7-012:0/16] 2024-12-06 02:55:24,610 (deepspeed_trainer:228) INFO: 17epoch:train:14001-14100batch: iter_time=1.107e-04, loss_ctc=78.176, loss_att=62.453, acc=0.692, loss=67.165, grad_norm=4.963, loss_scale=1.000, learning_rate=9.720e-05, step_time=0.364 [cnode7-012:0/16] 2024-12-06 02:56:00,965 (deepspeed_trainer:228) INFO: 17epoch:train:14101-14200batch: iter_time=1.155e-04, loss_ctc=77.532, loss_att=63.527, acc=0.673, loss=67.714, grad_norm=5.088, loss_scale=1.000, learning_rate=9.718e-05, step_time=0.363 [cnode7-012:0/16] 2024-12-06 02:56:37,144 (deepspeed_trainer:228) INFO: 17epoch:train:14201-14300batch: iter_time=1.243e-04, loss_ctc=75.304, loss_att=62.841, acc=0.687, loss=66.583, grad_norm=4.163, loss_scale=1.000, learning_rate=9.716e-05, step_time=0.361 [cnode7-012:0/16] 2024-12-06 02:57:13,348 (deepspeed_trainer:228) INFO: 17epoch:train:14301-14400batch: iter_time=1.157e-04, loss_ctc=68.571, loss_att=57.176, acc=0.681, loss=60.580, grad_norm=4.058, loss_scale=1.000, learning_rate=9.714e-05, step_time=0.362 [cnode7-012:0/16] 2024-12-06 02:57:49,853 (deepspeed_trainer:228) INFO: 17epoch:train:14401-14500batch: iter_time=1.089e-04, loss_ctc=71.622, loss_att=56.290, acc=0.701, loss=60.867, grad_norm=3.779, loss_scale=1.000, learning_rate=9.712e-05, step_time=0.365 [cnode7-012:0/16] 2024-12-06 02:58:26,827 (deepspeed_trainer:228) INFO: 17epoch:train:14501-14600batch: iter_time=1.120e-04, loss_ctc=82.364, loss_att=62.943, acc=0.686, loss=68.753, grad_norm=4.890, loss_scale=1.000, learning_rate=9.710e-05, step_time=0.369 [cnode7-012:0/16] 2024-12-06 02:59:03,031 (deepspeed_trainer:228) INFO: 17epoch:train:14601-14700batch: iter_time=1.140e-04, loss_ctc=70.856, loss_att=53.988, acc=0.693, loss=59.082, grad_norm=4.392, loss_scale=1.000, learning_rate=9.708e-05, step_time=0.362 [cnode7-012:0/16] 2024-12-06 02:59:39,123 (deepspeed_trainer:228) INFO: 17epoch:train:14701-14800batch: iter_time=1.117e-04, loss_ctc=76.683, loss_att=55.472, acc=0.698, loss=61.856, grad_norm=4.570, loss_scale=1.000, learning_rate=9.706e-05, step_time=0.361 [cnode7-012:0/16] 2024-12-06 03:00:15,315 (deepspeed_trainer:228) INFO: 17epoch:train:14801-14900batch: iter_time=1.096e-04, loss_ctc=75.164, loss_att=59.997, acc=0.694, loss=64.554, grad_norm=4.437, loss_scale=1.000, learning_rate=9.704e-05, step_time=0.362 [2024-12-06 03:00:51,272] [INFO] [logging.py:129:log_dist] [Rank 0] step=255000, skipped=0, lr=[np.float64(9.701405979107303e-05)], mom=[[0.9, 0.98]] [2024-12-06 03:00:51,273] [INFO] [timer.py:264:stop] epoch=0/micro_step=75000/global_step=75000, RunningAvgSamplesPerSec=44.975386177865445, CurrSamplesPerSec=48.47151037217861, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-06 03:00:51,274 (deepspeed_trainer:228) INFO: 17epoch:train:14901-15000batch: iter_time=1.079e-04, loss_ctc=70.638, loss_att=51.697, acc=0.690, loss=57.391, grad_norm=4.447, loss_scale=1.000, learning_rate=9.702e-05, step_time=0.359 [2024-12-06 03:01:05,388] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:06,063] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:07,642] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:06,652] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:08,245] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:07,143] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:07,202] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:07,228] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:07,259] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:07,337] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:08,910] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:08,916] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:09,008] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:09,010] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:09,092] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:09,103] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:21,075] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:23,217] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:21,782] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:22,198] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:24,322] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:23,615] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:24,080] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:24,281] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:24,365] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:24,403] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:25,925] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:26,271] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:26,415] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:26,568] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:26,805] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:27,001] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:35,956] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:38,607] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:37,415] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:38,355] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:39,945] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:39,375] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:39,831] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:41,329] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:41,650] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:40,354] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:40,445] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:40,482] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:42,031] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:42,323] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:42,824] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:42,941] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:50,882] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:53,586] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:52,717] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:55,350] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:54,127] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:56,436] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:55,185] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:55,245] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:56,839] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:55,489] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:57,389] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:56,070] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:56,107] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:57,830] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:58,390] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:01:58,424] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:02:11,153] [INFO] [logging.py:129:log_dist] [Rank 0] [Torch] Checkpoint 17 is about to be saved! [2024-12-06 03:02:11,186] [INFO] [logging.py:129:log_dist] [Rank 0] Saving model checkpoint: exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/mp_rank_00_model_states.pt [2024-12-06 03:02:11,186] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/mp_rank_00_model_states.pt... [2024-12-06 03:02:13,245] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/mp_rank_00_model_states.pt. [2024-12-06 03:02:13,402] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt... [2024-12-06 03:02:11,906] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_11_mp_rank_00_optim_states.pt... [2024-12-06 03:02:13,404] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt... [2024-12-06 03:02:13,405] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt... [2024-12-06 03:02:13,406] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt... [2024-12-06 03:02:13,406] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt... [2024-12-06 03:02:13,407] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt... [2024-12-06 03:02:11,911] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_15_mp_rank_00_optim_states.pt... [2024-12-06 03:02:11,911] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_14_mp_rank_00_optim_states.pt... [2024-12-06 03:02:11,913] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_10_mp_rank_00_optim_states.pt... [2024-12-06 03:02:11,913] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_12_mp_rank_00_optim_states.pt... [2024-12-06 03:02:13,410] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt... [2024-12-06 03:02:11,913] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_8_mp_rank_00_optim_states.pt... [2024-12-06 03:02:11,914] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_9_mp_rank_00_optim_states.pt... [2024-12-06 03:02:13,411] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt... [2024-12-06 03:02:11,914] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_13_mp_rank_00_optim_states.pt... [2024-12-06 03:02:14,083] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt. [2024-12-06 03:02:14,083] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt [2024-12-06 03:02:14,083] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 17 is ready now! [2024-12-06 03:02:14,126] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt. [2024-12-06 03:02:14,126] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt [2024-12-06 03:02:14,126] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 17 is ready now! [2024-12-06 03:02:14,146] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt. [2024-12-06 03:02:14,147] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt [2024-12-06 03:02:14,147] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 17 is ready now! [2024-12-06 03:02:12,680] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_8_mp_rank_00_optim_states.pt. [2024-12-06 03:02:12,680] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_8_mp_rank_00_optim_states.pt [2024-12-06 03:02:12,680] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 17 is ready now! [2024-12-06 03:02:12,691] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_11_mp_rank_00_optim_states.pt. [2024-12-06 03:02:12,691] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_11_mp_rank_00_optim_states.pt [2024-12-06 03:02:12,691] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 17 is ready now! [2024-12-06 03:02:14,190] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt. [2024-12-06 03:02:14,190] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt [2024-12-06 03:02:14,190] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 17 is ready now! [2024-12-06 03:02:12,695] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_9_mp_rank_00_optim_states.pt. [2024-12-06 03:02:12,695] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_9_mp_rank_00_optim_states.pt [2024-12-06 03:02:12,695] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 17 is ready now! [2024-12-06 03:02:12,702] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_15_mp_rank_00_optim_states.pt. [2024-12-06 03:02:12,702] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_15_mp_rank_00_optim_states.pt [2024-12-06 03:02:12,702] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 17 is ready now! [2024-12-06 03:02:12,714] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_13_mp_rank_00_optim_states.pt. [2024-12-06 03:02:12,715] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_13_mp_rank_00_optim_states.pt [2024-12-06 03:02:12,715] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 17 is ready now! [2024-12-06 03:02:14,227] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt. [2024-12-06 03:02:14,232] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt [2024-12-06 03:02:14,232] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 17 is ready now! [2024-12-06 03:02:12,738] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_12_mp_rank_00_optim_states.pt. [2024-12-06 03:02:12,738] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_12_mp_rank_00_optim_states.pt [2024-12-06 03:02:12,738] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 17 is ready now! [2024-12-06 03:02:12,738] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_10_mp_rank_00_optim_states.pt. [2024-12-06 03:02:12,739] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_10_mp_rank_00_optim_states.pt [2024-12-06 03:02:12,739] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 17 is ready now! [2024-12-06 03:02:14,249] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt. [2024-12-06 03:02:14,249] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt [2024-12-06 03:02:14,249] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 17 is ready now! [2024-12-06 03:02:12,753] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_14_mp_rank_00_optim_states.pt. [2024-12-06 03:02:12,754] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_14_mp_rank_00_optim_states.pt [2024-12-06 03:02:12,754] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 17 is ready now! [2024-12-06 03:02:14,253] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt. [2024-12-06 03:02:14,253] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt [2024-12-06 03:02:14,253] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 17 is ready now! [2024-12-06 03:02:14,255] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt. [2024-12-06 03:02:14,255] [INFO] [engine.py:3536:_save_zero_checkpoint] zero checkpoint saved exp_owsm/s2t_train_05b_ds_raw_bpe50000/checkpoint_17/17/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt [2024-12-06 03:02:14,255] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint 17 is ready now! [cnode7-012:0/16] 2024-12-06 03:02:14,264 (deepspeed_trainer:158) INFO: 17epoch results: [train] iter_time=0.134, loss_ctc=76.795, loss_att=60.163, acc=0.690, loss=65.153, grad_norm=4.492, loss_scale=1.000, learning_rate=9.848e-05, step_time=0.367, time=2 hours, 5 minutes and 32.74 seconds, total_count=255017, gpu_max_cached_mem_GB=36.545, [valid] loss_ctc=6.188, cer_ctc=0.196, loss_att=8.250, acc=0.775, cer=0.424, wer=1.000, loss=7.656, time=1 minute and 9.61 seconds, total_count=17, gpu_max_cached_mem_GB=36.545 [cnode7-012:0/16] 2024-12-06 03:02:16,020 (multiple_iter_factory:32) INFO: Building 0th iter-factory... [cnode7-012:0/16] 2024-12-06 03:02:42,145 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-06 03:02:58,171 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.4", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.4", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.4", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.4", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-06 03:02:58,171 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.4, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-06 03:02:58,174 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-06 03:03:18,079] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:03:18,102] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:03:16,688] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:03:17,366] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:03:17,865] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:03:18,359] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:03:18,768] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:03:20,568] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:03:20,768] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:03:19,279] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:03:19,316] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:03:19,580] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:03:21,117] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:03:21,895] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:03:22,254] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:03:22,685] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:04:04,601] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:04:06,968] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:04:06,028] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:04:07,786] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:04:06,536] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:04:06,966] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:04:07,022] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:04:07,344] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:04:07,600] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:04:07,868] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:04:11,376] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:04:12,126] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:04:12,739] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:04:12,808] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:04:12,824] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:04:12,861] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:04:52,445] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:04:54,232] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:04:52,794] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:04:55,150] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:04:54,076] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:04:54,656] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:04:55,172] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:04:55,262] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:04:58,617] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:04:58,273] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:04:58,707] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:05:00,323] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:05:01,096] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:05:01,521] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:05:01,555] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:05:04,502] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:05:42,329] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:05:42,522] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:05:41,308] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:05:41,698] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:05:42,404] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:05:42,878] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:05:43,154] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:05:44,155] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:05:47,860] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:05:48,270] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:05:47,433] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:05:47,804] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:05:50,294] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:05:50,536] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:05:50,964] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:05:55,986] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-06 03:06:59,789 (deepspeed_trainer:228) INFO: 18epoch:train:1-100batch: iter_time=2.396, loss_ctc=75.938, loss_att=58.458, acc=0.695, loss=63.713, grad_norm=4.344, loss_scale=1.000, learning_rate=9.700e-05, step_time=0.442 [cnode7-012:0/16] 2024-12-06 03:07:35,880 (deepspeed_trainer:228) INFO: 18epoch:train:101-200batch: iter_time=1.084e-04, loss_ctc=78.463, loss_att=57.364, acc=0.706, loss=63.687, grad_norm=4.428, loss_scale=1.000, learning_rate=9.699e-05, step_time=0.360 [cnode7-012:0/16] 2024-12-06 03:08:12,007 (deepspeed_trainer:228) INFO: 18epoch:train:201-300batch: iter_time=1.119e-04, loss_ctc=92.810, loss_att=71.016, acc=0.680, loss=77.530, grad_norm=5.625, loss_scale=1.000, learning_rate=9.697e-05, step_time=0.361 [cnode7-012:0/16] 2024-12-06 03:08:47,738 (deepspeed_trainer:228) INFO: 18epoch:train:301-400batch: iter_time=1.117e-04, loss_ctc=73.094, loss_att=53.575, acc=0.718, loss=59.455, grad_norm=4.236, loss_scale=1.000, learning_rate=9.695e-05, step_time=0.357 [cnode7-012:0/16] 2024-12-06 03:09:23,387 (deepspeed_trainer:228) INFO: 18epoch:train:401-500batch: iter_time=1.131e-04, loss_ctc=70.720, loss_att=51.868, acc=0.692, loss=57.527, grad_norm=4.013, loss_scale=1.000, learning_rate=9.693e-05, step_time=0.356 [cnode7-012:0/16] 2024-12-06 03:09:59,086 (deepspeed_trainer:228) INFO: 18epoch:train:501-600batch: iter_time=1.133e-04, loss_ctc=72.676, loss_att=60.222, acc=0.694, loss=63.960, grad_norm=4.212, loss_scale=1.000, learning_rate=9.691e-05, step_time=0.357 [cnode7-012:0/16] 2024-12-06 03:10:34,853 (deepspeed_trainer:228) INFO: 18epoch:train:601-700batch: iter_time=1.120e-04, loss_ctc=77.244, loss_att=56.617, acc=0.702, loss=62.836, grad_norm=4.794, loss_scale=1.000, learning_rate=9.689e-05, step_time=0.357 [cnode7-012:0/16] 2024-12-06 03:11:10,642 (deepspeed_trainer:228) INFO: 18epoch:train:701-800batch: iter_time=1.155e-04, loss_ctc=75.651, loss_att=59.798, acc=0.689, loss=64.603, grad_norm=4.807, loss_scale=1.000, learning_rate=9.687e-05, step_time=0.357 [cnode7-012:0/16] 2024-12-06 03:11:46,392 (deepspeed_trainer:228) INFO: 18epoch:train:801-900batch: iter_time=1.176e-04, loss_ctc=76.791, loss_att=55.879, acc=0.717, loss=62.150, grad_norm=3.812, loss_scale=1.000, learning_rate=9.685e-05, step_time=0.357 [2024-12-06 03:12:22,563] [INFO] [logging.py:129:log_dist] [Rank 0] step=256000, skipped=0, lr=[np.float64(9.682439454522451e-05)], mom=[[0.9, 0.98]] [2024-12-06 03:12:22,563] [INFO] [timer.py:264:stop] epoch=0/micro_step=76000/global_step=76000, RunningAvgSamplesPerSec=44.973316519258496, CurrSamplesPerSec=47.12889128474097, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-06 03:12:22,565 (deepspeed_trainer:228) INFO: 18epoch:train:901-1000batch: iter_time=1.215e-04, loss_ctc=92.047, loss_att=69.532, acc=0.680, loss=76.287, grad_norm=5.948, loss_scale=1.000, learning_rate=9.683e-05, step_time=0.362 [cnode7-012:0/16] 2024-12-06 03:12:59,009 (deepspeed_trainer:228) INFO: 18epoch:train:1001-1100batch: iter_time=1.216e-04, loss_ctc=67.793, loss_att=54.530, acc=0.698, loss=58.492, grad_norm=4.237, loss_scale=1.000, learning_rate=9.681e-05, step_time=0.364 [cnode7-012:0/16] 2024-12-06 03:13:34,968 (deepspeed_trainer:228) INFO: 18epoch:train:1101-1200batch: iter_time=1.122e-04, loss_ctc=66.182, loss_att=49.681, acc=0.701, loss=54.648, grad_norm=4.098, loss_scale=1.000, learning_rate=9.680e-05, step_time=0.359 [cnode7-012:0/16] 2024-12-06 03:14:11,245 (deepspeed_trainer:228) INFO: 18epoch:train:1201-1300batch: iter_time=1.132e-04, loss_ctc=74.921, loss_att=57.486, acc=0.692, loss=62.709, grad_norm=4.433, loss_scale=1.000, learning_rate=9.678e-05, step_time=0.363 [cnode7-012:0/16] 2024-12-06 03:14:47,524 (deepspeed_trainer:228) INFO: 18epoch:train:1301-1400batch: iter_time=1.257e-04, loss_ctc=78.408, loss_att=55.603, acc=0.691, loss=62.449, grad_norm=4.697, loss_scale=1.000, learning_rate=9.676e-05, step_time=0.362 [cnode7-012:0/16] 2024-12-06 03:15:24,235 (deepspeed_trainer:228) INFO: 18epoch:train:1401-1500batch: iter_time=1.132e-04, loss_ctc=82.600, loss_att=73.251, acc=0.672, loss=76.052, grad_norm=4.654, loss_scale=1.000, learning_rate=9.674e-05, step_time=0.367 [cnode7-012:0/16] 2024-12-06 03:16:00,749 (deepspeed_trainer:228) INFO: 18epoch:train:1501-1600batch: iter_time=1.109e-04, loss_ctc=74.936, loss_att=61.330, acc=0.700, loss=65.388, grad_norm=4.164, loss_scale=1.000, learning_rate=9.672e-05, step_time=0.365 [cnode7-012:0/16] 2024-12-06 03:16:37,081 (deepspeed_trainer:228) INFO: 18epoch:train:1601-1700batch: iter_time=1.123e-04, loss_ctc=80.647, loss_att=57.284, acc=0.712, loss=64.306, grad_norm=4.415, loss_scale=1.000, learning_rate=9.670e-05, step_time=0.363 [cnode7-012:0/16] 2024-12-06 03:17:13,013 (deepspeed_trainer:228) INFO: 18epoch:train:1701-1800batch: iter_time=1.120e-04, loss_ctc=78.904, loss_att=64.137, acc=0.690, loss=68.561, grad_norm=4.486, loss_scale=1.000, learning_rate=9.668e-05, step_time=0.358 [cnode7-012:0/16] 2024-12-06 03:17:43,753 (multiple_iter_factory:32) INFO: Building 1th iter-factory... [cnode7-012:0/16] 2024-12-06 03:18:10,867 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-06 03:18:27,560 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.7", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.7", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.7", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.7", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-06 03:18:27,560 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.7, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-06 03:18:27,562 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-06 03:18:55,543] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:18:55,585] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:18:56,060] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:18:58,709] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:18:57,254] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:18:57,259] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:18:57,464] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:18:57,575] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:18:57,593] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:18:57,697] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:18:57,711] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:18:57,724] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:18:59,484] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:18:59,854] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:19:00,215] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:19:00,266] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:19:46,374] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:19:46,411] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:19:47,006] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:19:49,062] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:19:49,384] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:19:50,140] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:19:51,330] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:19:51,352] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:19:51,755] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:19:51,806] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:19:51,996] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:19:52,121] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:19:52,127] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:19:52,157] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:19:52,163] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:19:53,354] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:20:35,278] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:20:35,562] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:20:36,211] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:20:39,290] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:20:39,425] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:20:39,730] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:20:40,554] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:20:40,892] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:20:40,271] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:20:40,576] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:20:40,816] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:20:40,991] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:20:41,465] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:20:42,068] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:20:44,091] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:20:44,378] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:21:22,540] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:21:22,897] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:21:24,346] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:21:27,799] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:21:28,673] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:21:28,980] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:21:29,881] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:21:28,679] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:21:30,333] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:21:28,815] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:21:29,133] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:21:29,698] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:21:29,769] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:21:30,784] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:21:35,140] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:21:35,736] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-06 03:22:13,235 (deepspeed_trainer:228) INFO: 18epoch:train:1801-1900batch: iter_time=2.514, loss_ctc=89.673, loss_att=66.444, acc=0.696, loss=73.402, grad_norm=5.504, loss_scale=1.000, learning_rate=9.666e-05, step_time=0.488 [2024-12-06 03:22:49,014] [INFO] [logging.py:129:log_dist] [Rank 0] step=257000, skipped=0, lr=[np.float64(9.663583737030714e-05)], mom=[[0.9, 0.98]] [2024-12-06 03:22:49,017] [INFO] [timer.py:264:stop] epoch=0/micro_step=77000/global_step=77000, RunningAvgSamplesPerSec=44.95841604415347, CurrSamplesPerSec=47.82292226331128, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-06 03:22:49,019 (deepspeed_trainer:228) INFO: 18epoch:train:1901-2000batch: iter_time=1.064e-04, loss_ctc=74.444, loss_att=53.827, acc=0.699, loss=60.009, grad_norm=4.319, loss_scale=1.000, learning_rate=9.665e-05, step_time=0.358 [cnode7-012:0/16] 2024-12-06 03:23:24,892 (deepspeed_trainer:228) INFO: 18epoch:train:2001-2100batch: iter_time=1.129e-04, loss_ctc=87.374, loss_att=65.394, acc=0.694, loss=71.965, grad_norm=4.947, loss_scale=1.000, learning_rate=9.663e-05, step_time=0.358 [cnode7-012:0/16] 2024-12-06 03:24:00,858 (deepspeed_trainer:228) INFO: 18epoch:train:2101-2200batch: iter_time=1.103e-04, loss_ctc=75.767, loss_att=57.604, acc=0.691, loss=63.068, grad_norm=4.523, loss_scale=1.000, learning_rate=9.661e-05, step_time=0.359 [cnode7-012:0/16] 2024-12-06 03:24:37,013 (deepspeed_trainer:228) INFO: 18epoch:train:2201-2300batch: iter_time=1.128e-04, loss_ctc=71.847, loss_att=51.356, acc=0.712, loss=57.509, grad_norm=4.093, loss_scale=1.000, learning_rate=9.659e-05, step_time=0.361 [cnode7-012:0/16] 2024-12-06 03:25:13,268 (deepspeed_trainer:228) INFO: 18epoch:train:2301-2400batch: iter_time=1.115e-04, loss_ctc=71.597, loss_att=52.707, acc=0.693, loss=58.354, grad_norm=4.122, loss_scale=1.000, learning_rate=9.657e-05, step_time=0.362 [cnode7-012:0/16] 2024-12-06 03:25:49,611 (deepspeed_trainer:228) INFO: 18epoch:train:2401-2500batch: iter_time=1.093e-04, loss_ctc=74.412, loss_att=61.783, acc=0.687, loss=65.564, grad_norm=4.203, loss_scale=1.000, learning_rate=9.655e-05, step_time=0.363 [cnode7-012:0/16] 2024-12-06 03:26:25,746 (deepspeed_trainer:228) INFO: 18epoch:train:2501-2600batch: iter_time=1.105e-04, loss_ctc=76.469, loss_att=56.565, acc=0.693, loss=62.504, grad_norm=4.522, loss_scale=1.000, learning_rate=9.653e-05, step_time=0.361 [cnode7-012:0/16] 2024-12-06 03:27:01,794 (deepspeed_trainer:228) INFO: 18epoch:train:2601-2700batch: iter_time=1.087e-04, loss_ctc=76.855, loss_att=58.928, acc=0.696, loss=64.312, grad_norm=4.453, loss_scale=1.000, learning_rate=9.651e-05, step_time=0.360 [cnode7-012:0/16] 2024-12-06 03:27:37,686 (deepspeed_trainer:228) INFO: 18epoch:train:2701-2800batch: iter_time=1.089e-04, loss_ctc=79.046, loss_att=59.638, acc=0.712, loss=65.443, grad_norm=4.546, loss_scale=1.000, learning_rate=9.650e-05, step_time=0.359 [cnode7-012:0/16] 2024-12-06 03:28:14,118 (deepspeed_trainer:228) INFO: 18epoch:train:2801-2900batch: iter_time=1.097e-04, loss_ctc=84.942, loss_att=66.045, acc=0.676, loss=71.674, grad_norm=4.893, loss_scale=1.000, learning_rate=9.648e-05, step_time=0.364 [2024-12-06 03:28:49,609] [INFO] [logging.py:129:log_dist] [Rank 0] step=258000, skipped=0, lr=[np.float64(9.644837751880323e-05)], mom=[[0.9, 0.98]] [2024-12-06 03:28:49,610] [INFO] [timer.py:264:stop] epoch=0/micro_step=78000/global_step=78000, RunningAvgSamplesPerSec=44.96708708732106, CurrSamplesPerSec=47.49657076920584, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-06 03:28:49,612 (deepspeed_trainer:228) INFO: 18epoch:train:2901-3000batch: iter_time=1.091e-04, loss_ctc=65.183, loss_att=48.997, acc=0.705, loss=53.854, grad_norm=4.078, loss_scale=1.000, learning_rate=9.646e-05, step_time=0.355 [cnode7-012:0/16] 2024-12-06 03:29:25,087 (deepspeed_trainer:228) INFO: 18epoch:train:3001-3100batch: iter_time=1.071e-04, loss_ctc=62.656, loss_att=45.691, acc=0.699, loss=50.767, grad_norm=3.972, loss_scale=1.000, learning_rate=9.644e-05, step_time=0.354 [cnode7-012:0/16] 2024-12-06 03:30:00,821 (deepspeed_trainer:228) INFO: 18epoch:train:3101-3200batch: iter_time=1.087e-04, loss_ctc=77.728, loss_att=58.433, acc=0.697, loss=64.209, grad_norm=4.386, loss_scale=1.000, learning_rate=9.642e-05, step_time=0.357 [cnode7-012:0/16] 2024-12-06 03:30:36,717 (deepspeed_trainer:228) INFO: 18epoch:train:3201-3300batch: iter_time=1.103e-04, loss_ctc=78.584, loss_att=59.415, acc=0.687, loss=65.162, grad_norm=4.750, loss_scale=1.000, learning_rate=9.640e-05, step_time=0.359 [cnode7-012:0/16] 2024-12-06 03:31:12,657 (deepspeed_trainer:228) INFO: 18epoch:train:3301-3400batch: iter_time=1.079e-04, loss_ctc=81.881, loss_att=69.478, acc=0.671, loss=73.179, grad_norm=4.619, loss_scale=1.000, learning_rate=9.638e-05, step_time=0.359 [cnode7-012:0/16] 2024-12-06 03:31:48,375 (deepspeed_trainer:228) INFO: 18epoch:train:3401-3500batch: iter_time=1.090e-04, loss_ctc=71.834, loss_att=56.515, acc=0.699, loss=61.127, grad_norm=4.358, loss_scale=1.000, learning_rate=9.636e-05, step_time=0.357 [cnode7-012:0/16] 2024-12-06 03:32:24,279 (deepspeed_trainer:228) INFO: 18epoch:train:3501-3600batch: iter_time=1.102e-04, loss_ctc=84.704, loss_att=65.389, acc=0.694, loss=71.175, grad_norm=4.303, loss_scale=1.000, learning_rate=9.635e-05, step_time=0.359 [cnode7-012:0/16] 2024-12-06 03:33:00,061 (deepspeed_trainer:228) INFO: 18epoch:train:3601-3700batch: iter_time=1.104e-04, loss_ctc=75.927, loss_att=55.740, acc=0.700, loss=61.827, grad_norm=4.397, loss_scale=1.000, learning_rate=9.633e-05, step_time=0.357 [cnode7-012:0/16] 2024-12-06 03:33:20,655 (multiple_iter_factory:32) INFO: Building 2th iter-factory... [cnode7-012:0/16] 2024-12-06 03:33:47,548 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-06 03:34:02,977 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.6", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.6", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.6", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.6", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-06 03:34:02,977 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.6, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-06 03:34:02,979 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-06 03:34:29,232] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:34:29,720] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:34:30,500] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:34:33,449] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:34:32,049] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:34:32,144] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:34:33,784] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:34:34,186] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:34:32,701] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:34:32,946] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:34:34,778] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:34:33,249] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:34:33,260] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:34:33,288] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:34:33,445] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:34:35,001] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:35:17,274] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:35:18,386] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:35:19,690] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:35:20,608] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:35:20,920] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:35:22,478] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:35:21,489] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:35:23,279] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:35:21,789] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:35:22,118] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:35:24,044] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:35:22,523] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:35:22,577] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:35:22,719] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:35:24,361] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:35:32,298] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:36:06,744] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:36:07,028] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:36:07,906] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:36:11,662] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:36:10,161] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:36:10,446] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:36:10,697] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:36:12,269] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:36:10,749] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:36:11,439] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:36:11,453] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:36:11,618] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:36:13,667] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:36:13,892] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:36:16,755] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:36:25,930] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:36:53,623] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:36:55,159] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:36:55,215] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:36:57,603] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:36:58,316] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:36:58,536] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:37:00,472] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:36:58,986] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:36:59,334] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:37:01,303] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:36:59,886] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:37:00,114] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:37:01,675] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:37:04,599] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:37:07,068] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:37:19,822] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-06 03:38:03,258 (deepspeed_trainer:228) INFO: 18epoch:train:3701-3800batch: iter_time=2.444, loss_ctc=89.462, loss_att=67.704, acc=0.688, loss=74.232, grad_norm=4.842, loss_scale=1.000, learning_rate=9.631e-05, step_time=0.587 [cnode7-012:0/16] 2024-12-06 03:38:39,768 (deepspeed_trainer:228) INFO: 18epoch:train:3801-3900batch: iter_time=1.097e-04, loss_ctc=70.271, loss_att=53.052, acc=0.714, loss=58.227, grad_norm=3.736, loss_scale=1.000, learning_rate=9.629e-05, step_time=0.365 [2024-12-06 03:39:16,375] [INFO] [logging.py:129:log_dist] [Rank 0] step=259000, skipped=0, lr=[np.float64(9.626200438857197e-05)], mom=[[0.9, 0.98]] [2024-12-06 03:39:16,376] [INFO] [timer.py:264:stop] epoch=0/micro_step=79000/global_step=79000, RunningAvgSamplesPerSec=44.94061114565171, CurrSamplesPerSec=43.073388157901476, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-06 03:39:16,378 (deepspeed_trainer:228) INFO: 18epoch:train:3901-4000batch: iter_time=1.132e-04, loss_ctc=89.391, loss_att=68.618, acc=0.687, loss=74.881, grad_norm=5.182, loss_scale=1.000, learning_rate=9.627e-05, step_time=0.366 [cnode7-012:0/16] 2024-12-06 03:39:52,780 (deepspeed_trainer:228) INFO: 18epoch:train:4001-4100batch: iter_time=1.179e-04, loss_ctc=75.267, loss_att=57.372, acc=0.701, loss=62.709, grad_norm=4.394, loss_scale=1.000, learning_rate=9.625e-05, step_time=0.364 [cnode7-012:0/16] 2024-12-06 03:40:28,840 (deepspeed_trainer:228) INFO: 18epoch:train:4101-4200batch: iter_time=1.203e-04, loss_ctc=71.288, loss_att=51.095, acc=0.717, loss=57.186, grad_norm=4.114, loss_scale=1.000, learning_rate=9.623e-05, step_time=0.360 [cnode7-012:0/16] 2024-12-06 03:41:05,075 (deepspeed_trainer:228) INFO: 18epoch:train:4201-4300batch: iter_time=1.165e-04, loss_ctc=68.131, loss_att=54.324, acc=0.695, loss=58.449, grad_norm=4.332, loss_scale=1.000, learning_rate=9.622e-05, step_time=0.361 [cnode7-012:0/16] 2024-12-06 03:41:42,236 (deepspeed_trainer:228) INFO: 18epoch:train:4301-4400batch: iter_time=1.207e-04, loss_ctc=77.528, loss_att=62.170, acc=0.695, loss=66.762, grad_norm=4.632, loss_scale=1.000, learning_rate=9.620e-05, step_time=0.372 [cnode7-012:0/16] 2024-12-06 03:42:18,643 (deepspeed_trainer:228) INFO: 18epoch:train:4401-4500batch: iter_time=1.177e-04, loss_ctc=72.814, loss_att=57.897, acc=0.699, loss=62.361, grad_norm=4.268, loss_scale=1.000, learning_rate=9.618e-05, step_time=0.363 [cnode7-012:0/16] 2024-12-06 03:42:54,939 (deepspeed_trainer:228) INFO: 18epoch:train:4501-4600batch: iter_time=1.126e-04, loss_ctc=73.953, loss_att=53.861, acc=0.710, loss=59.886, grad_norm=4.254, loss_scale=1.000, learning_rate=9.616e-05, step_time=0.363 [cnode7-012:0/16] 2024-12-06 03:43:31,449 (deepspeed_trainer:228) INFO: 18epoch:train:4601-4700batch: iter_time=1.153e-04, loss_ctc=85.023, loss_att=66.864, acc=0.700, loss=72.327, grad_norm=4.925, loss_scale=1.000, learning_rate=9.614e-05, step_time=0.364 [cnode7-012:0/16] 2024-12-06 03:44:07,466 (deepspeed_trainer:228) INFO: 18epoch:train:4701-4800batch: iter_time=1.138e-04, loss_ctc=75.692, loss_att=61.126, acc=0.698, loss=65.495, grad_norm=4.402, loss_scale=1.000, learning_rate=9.612e-05, step_time=0.360 [cnode7-012:0/16] 2024-12-06 03:44:43,161 (deepspeed_trainer:228) INFO: 18epoch:train:4801-4900batch: iter_time=1.161e-04, loss_ctc=63.894, loss_att=47.178, acc=0.704, loss=52.201, grad_norm=4.210, loss_scale=1.000, learning_rate=9.610e-05, step_time=0.356 [2024-12-06 03:45:18,621] [INFO] [logging.py:129:log_dist] [Rank 0] step=260000, skipped=0, lr=[np.float64(9.607670752033088e-05)], mom=[[0.9, 0.98]] [2024-12-06 03:45:18,621] [INFO] [timer.py:264:stop] epoch=0/micro_step=80000/global_step=80000, RunningAvgSamplesPerSec=44.94693736261438, CurrSamplesPerSec=44.50286733921556, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-06 03:45:18,624 (deepspeed_trainer:228) INFO: 18epoch:train:4901-5000batch: iter_time=1.144e-04, loss_ctc=69.101, loss_att=49.687, acc=0.701, loss=55.509, grad_norm=4.394, loss_scale=1.000, learning_rate=9.609e-05, step_time=0.354 [cnode7-012:0/16] 2024-12-06 03:45:54,751 (deepspeed_trainer:228) INFO: 18epoch:train:5001-5100batch: iter_time=1.175e-04, loss_ctc=75.956, loss_att=57.538, acc=0.704, loss=63.053, grad_norm=4.741, loss_scale=1.000, learning_rate=9.607e-05, step_time=0.361 [cnode7-012:0/16] 2024-12-06 03:46:30,829 (deepspeed_trainer:228) INFO: 18epoch:train:5101-5200batch: iter_time=1.151e-04, loss_ctc=78.471, loss_att=64.975, acc=0.678, loss=69.016, grad_norm=4.602, loss_scale=1.000, learning_rate=9.605e-05, step_time=0.360 [cnode7-012:0/16] 2024-12-06 03:47:06,837 (deepspeed_trainer:228) INFO: 18epoch:train:5201-5300batch: iter_time=1.187e-04, loss_ctc=74.594, loss_att=62.188, acc=0.694, loss=65.878, grad_norm=4.488, loss_scale=1.000, learning_rate=9.603e-05, step_time=0.359 [cnode7-012:0/16] 2024-12-06 03:47:42,708 (deepspeed_trainer:228) INFO: 18epoch:train:5301-5400batch: iter_time=1.128e-04, loss_ctc=78.617, loss_att=59.538, acc=0.703, loss=65.284, grad_norm=4.371, loss_scale=1.000, learning_rate=9.601e-05, step_time=0.358 [cnode7-012:0/16] 2024-12-06 03:48:18,774 (deepspeed_trainer:228) INFO: 18epoch:train:5401-5500batch: iter_time=1.143e-04, loss_ctc=78.810, loss_att=63.027, acc=0.695, loss=67.769, grad_norm=4.941, loss_scale=1.000, learning_rate=9.599e-05, step_time=0.361 [cnode7-012:0/16] 2024-12-06 03:48:55,290 (deepspeed_trainer:228) INFO: 18epoch:train:5501-5600batch: iter_time=1.138e-04, loss_ctc=87.012, loss_att=62.119, acc=0.704, loss=69.628, grad_norm=4.949, loss_scale=1.000, learning_rate=9.598e-05, step_time=0.365 [cnode7-012:0/16] 2024-12-06 03:49:07,100 (multiple_iter_factory:32) INFO: Building 3th iter-factory... [cnode7-012:0/16] 2024-12-06 03:49:33,792 (s2t:444) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4') [cnode7-012:0/16] 2024-12-06 03:49:50,236 (abs_task:1807) INFO: [train] dataset: ESPnetDataset( speech: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/wav.scp/split.0", "type": "kaldi_ark"} text_prev: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.prev/split.0", "type": "text"} text_ctc: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text.ctc/split.0", "type": "text"} text: {"path": "exp_owsm/s2t_stats_raw_bpe50000/splits8/text/split.0", "type": "text"} preprocess: ) [cnode7-012:0/16] 2024-12-06 03:49:50,237 (abs_task:1808) INFO: [train] Batch sampler: SortedBatchSampler(N-batch=28521, batch_size=256, shape_file=exp_owsm/s2t_stats_raw_bpe50000/splits8/speech_shape/split.0, sort_in_batch=descending, sort_batch=descending) [cnode7-012:0/16] 2024-12-06 03:49:50,239 (abs_task:1809) INFO: [train] mini-batch sizes summary: N-batch=28521, mean=256.0, min=256, max=257 [2024-12-06 03:50:16,657] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:50:17,075] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:50:15,526] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:50:16,427] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:50:18,124] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:50:16,761] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:50:17,204] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:50:17,533] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:50:18,086] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:50:18,240] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:50:18,534] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:50:20,607] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:50:21,041] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:50:21,355] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:50:21,963] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:50:22,361] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:51:04,818] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:51:05,146] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:51:04,154] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:51:06,466] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:51:05,169] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:51:05,755] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:51:05,983] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:51:06,442] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:51:07,056] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:51:07,346] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:51:10,795] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:51:11,095] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:51:10,028] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:51:11,605] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:51:11,647] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:51:12,259] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:51:52,867] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:51:53,769] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:51:53,215] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:51:54,792] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:51:53,738] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:51:55,158] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:51:55,398] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:51:55,833] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:51:55,875] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:51:56,071] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:51:59,726] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:52:00,961] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:52:01,018] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:52:01,636] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:52:01,681] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:52:01,563] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:52:40,489] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:52:41,306] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:52:41,143] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:52:43,325] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:52:43,375] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:52:43,697] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:52:44,052] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:52:44,796] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:52:46,102] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:52:47,124] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:52:48,806] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:52:50,164] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:52:50,871] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:52:50,951] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:52:50,996] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2024-12-06 03:52:51,238] [INFO] [real_accelerator.py:219:get_accelerator] Setting ds_accelerator to cuda (auto detect) [cnode7-012:0/16] 2024-12-06 03:53:46,082 (deepspeed_trainer:228) INFO: 18epoch:train:5601-5700batch: iter_time=2.460, loss_ctc=80.301, loss_att=62.171, acc=0.701, loss=67.586, grad_norm=4.375, loss_scale=1.000, learning_rate=9.596e-05, step_time=0.448 [cnode7-012:0/16] 2024-12-06 03:54:22,343 (deepspeed_trainer:228) INFO: 18epoch:train:5701-5800batch: iter_time=1.102e-04, loss_ctc=69.171, loss_att=52.045, acc=0.711, loss=57.185, grad_norm=3.942, loss_scale=1.000, learning_rate=9.594e-05, step_time=0.362 [cnode7-012:0/16] 2024-12-06 03:54:58,760 (deepspeed_trainer:228) INFO: 18epoch:train:5801-5900batch: iter_time=1.153e-04, loss_ctc=88.423, loss_att=67.233, acc=0.693, loss=73.560, grad_norm=5.142, loss_scale=1.000, learning_rate=9.592e-05, step_time=0.364 [2024-12-06 03:55:34,827] [INFO] [logging.py:129:log_dist] [Rank 0] step=261000, skipped=0, lr=[np.float64(9.589247659519035e-05)], mom=[[0.9, 0.98]] [2024-12-06 03:55:34,828] [INFO] [timer.py:264:stop] epoch=0/micro_step=81000/global_step=81000, RunningAvgSamplesPerSec=44.940720917261835, CurrSamplesPerSec=43.99095089355548, MemAllocated=2.04GB, MaxMemAllocated=29.88GB [cnode7-012:0/16] 2024-12-06 03:55:34,830 (deepspeed_trainer:228) INFO: 18epoch:train:5901-6000batch: iter_time=1.211e-04, loss_ctc=76.404, loss_att=55.798, acc=0.712, loss=61.961, grad_norm=4.711, loss_scale=1.000, learning_rate=9.590e-05, step_time=0.360 [cnode7-012:0/16] 2024-12-06 03:56:10,450 (deepspeed_trainer:228) INFO: 18epoch:train:6001-6100batch: iter_time=1.166e-04, loss_ctc=67.242, loss_att=49.557, acc=0.712, loss=54.874, grad_norm=3.762, loss_scale=1.000, learning_rate=9.588e-05, step_time=0.356 [cnode7-012:0/16] 2024-12-06 03:56:46,832 (deepspeed_trainer:228) INFO: 18epoch:train:6101-6200batch: iter_time=1.210e-04, loss_ctc=70.517, loss_att=59.312, acc=0.696, loss=62.672, grad_norm=4.304, loss_scale=1.000, learning_rate=9.586e-05, step_time=0.363 [cnode7-012:0/16] 2024-12-06 03:57:22,772 (deepspeed_trainer:228) INFO: 18epoch:train:6201-6300batch: iter_time=1.254e-04, loss_ctc=78.208, loss_att=57.698, acc=0.703, loss=63.875, grad_norm=4.556, loss_scale=1.000, learning_rate=9.585e-05, step_time=0.359 [cnode7-012:0/16] 2024-12-06 03:57:58,657 (deepspeed_trainer:228) INFO: 18epoch:train:6301-6400batch: iter_time=1.204e-04, loss_ctc=74.114, loss_att=58.680, acc=0.702, loss=63.340, grad_norm=4.683, loss_scale=1.000, learning_rate=9.583e-05, step_time=0.359 [cnode7-012:0/16] 2024-12-06 03:58:34,433 (deepspeed_trainer:228) INFO: 18epoch:train:6401-6500batch: iter_time=1.155e-04, loss_ctc=74.979, loss_att=54.363, acc=0.715, loss=60.572, grad_norm=4.306, loss_scale=1.000, learning_rate=9.581e-05, step_time=0.357 [cnode7-012:0/16] 2024-12-06 03:59:10,597 (deepspeed_trainer:228) INFO: 18epoch:train:6501-6600batch: iter_time=1.126e-04, loss_ctc=81.801, loss_att=64.707, acc=0.690, loss=69.856, grad_norm=5.289, loss_scale=1.000, learning_rate=9.579e-05, step_time=0.361 Process SpawnProcess-6: Traceback (most recent call last): File "/mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/multiprocessing/process.py", line 314, in _bootstrap self.run() File "/mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/multiprocessing/process.py", line 108, in run self._target(*self._args, **self._kwargs) File "/mnt/home/williamchen/espnet/espnet2/tasks/abs_task.py", line 1568, in main_worker cls.trainer.run( File "/mnt/home/williamchen/espnet/espnet2/train/deepspeed_trainer.py", line 132, in run cls.train_one_epoch( File "/mnt/home/williamchen/espnet/espnet2/train/deepspeed_trainer.py", line 204, in train_one_epoch loss, stats, weight = model(**batch) File "/mnt/home/williamchen/.local/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1553, in _wrapped_call_impl return self._call_impl(*args, **kwargs) File "/mnt/home/williamchen/.local/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1562, in _call_impl return forward_call(*args, **kwargs) File "/mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/site-packages/deepspeed/utils/nvtx.py", line 18, in wrapped_fn ret_val = func(*args, **kwargs) File "/mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/site-packages/deepspeed/runtime/engine.py", line 1899, in forward loss = self.module(*inputs, **kwargs) File "/mnt/home/williamchen/.local/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1553, in _wrapped_call_impl return self._call_impl(*args, **kwargs) File "/mnt/home/williamchen/.local/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1562, in _call_impl return forward_call(*args, **kwargs) File "/mnt/home/williamchen/espnet/espnet2/s2t/espnet_model.py", line 225, in forward loss_att, acc_att, cer_att, wer_att = self._calc_att_loss( File "/mnt/home/williamchen/espnet/espnet2/s2t/espnet_model.py", line 391, in _calc_att_loss decoder_out, _ = self.decoder( File "/mnt/home/williamchen/.local/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1553, in _wrapped_call_impl return self._call_impl(*args, **kwargs) File "/mnt/home/williamchen/.local/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1562, in _call_impl return forward_call(*args, **kwargs) File "/mnt/home/williamchen/espnet/espnet2/asr/decoder/transformer_decoder.py", line 155, in forward x, tgt_mask, memory, memory_mask = checkpoint(decoder_layer, File "/mnt/home/williamchen/.local/lib/python3.10/site-packages/torch/_dynamo/eval_frame.py", line 600, in _fn return fn(*args, **kwargs) File "/mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/site-packages/deepspeed/runtime/activation_checkpointing/checkpointing.py", line 951, in checkpoint CheckpointFunction.apply(function, all_outputs, *args) File "/mnt/home/williamchen/.local/lib/python3.10/site-packages/torch/autograd/function.py", line 574, in apply return super().apply(*args, **kwargs) # type: ignore[misc] File "/mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/site-packages/deepspeed/runtime/activation_checkpointing/checkpointing.py", line 542, in forward outputs = run_function(*inputs_cuda) File "/mnt/home/williamchen/.local/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1553, in _wrapped_call_impl return self._call_impl(*args, **kwargs) File "/mnt/home/williamchen/.local/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1562, in _call_impl return forward_call(*args, **kwargs) File "/mnt/home/williamchen/espnet/espnet/nets/pytorch_backend/transformer/decoder_layer.py", line 170, in forward x = residual + self.dropout(self.feed_forward(x)) File "/mnt/home/williamchen/.local/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1553, in _wrapped_call_impl return self._call_impl(*args, **kwargs) File "/mnt/home/williamchen/.local/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1562, in _call_impl return forward_call(*args, **kwargs) File "/mnt/home/williamchen/espnet/espnet/nets/pytorch_backend/transformer/positionwise_feed_forward.py", line 32, in forward return self.w_2(self.dropout(self.activation(self.w_1(x)))) File "/mnt/home/williamchen/.local/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1553, in _wrapped_call_impl return self._call_impl(*args, **kwargs) File "/mnt/home/williamchen/.local/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1562, in _call_impl return forward_call(*args, **kwargs) File "/mnt/home/williamchen/.local/lib/python3.10/site-packages/torch/nn/modules/linear.py", line 117, in forward return F.linear(input, self.weight, self.bias) torch.OutOfMemoryError: CUDA out of memory. Tried to allocate 6.91 GiB. GPU 5 has a total capacity of 79.11 GiB of which 5.42 GiB is free. Including non-PyTorch memory, this process has 73.68 GiB memory in use. Of the allocated memory 67.16 GiB is allocated by PyTorch, and 462.60 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True to avoid fragmentation. See documentation for Memory Management (https://pytorch.org/docs/stable/notes/cuda.html#environment-variables) cnode7-012:1272005:1281945 [0] NCCL INFO [Service thread] Connection closed by localRank 5 cnode7-012:1272011:1281936 [6] NCCL INFO [Service thread] Connection closed by localRank 5 cnode7-012:1272011:1273093 [6] NCCL INFO [Service thread] Connection closed by localRank 5 cnode7-012:1272009:1281935 [4] NCCL INFO [Service thread] Connection closed by localRank 5 cnode7-012:1272009:1273094 [4] NCCL INFO [Service thread] Connection closed by localRank 5 cnode7-012:1272007:1281931 [2] NCCL INFO [Service thread] Connection closed by localRank 5 cnode7-012:1272007:1273091 [2] NCCL INFO [Service thread] Connection closed by localRank 5 cnode7-012:1272008:1281937 [3] NCCL INFO [Service thread] Connection closed by localRank 5 cnode7-012:1272008:1273095 [3] NCCL INFO [Service thread] Connection closed by localRank 5 cnode7-012:1272006:1281938 [1] NCCL INFO [Service thread] Connection closed by localRank 5 cnode7-012:1272006:1273097 [1] NCCL INFO [Service thread] Connection closed by localRank 5 cnode7-012:1272012:1281933 [7] NCCL INFO [Service thread] Connection closed by localRank 5 cnode7-012:1272012:1273092 [7] NCCL INFO [Service thread] Connection closed by localRank 5 cnode7-012:1272005:1273104 [0] NCCL INFO [Service thread] Connection closed by localRank 5 W1206 03:59:25.713000 23456244418368 torch/multiprocessing/spawn.py:146] Terminating process 1272005 via signal SIGTERM W1206 03:59:25.714000 23456244418368 torch/multiprocessing/spawn.py:146] Terminating process 1272006 via signal SIGTERM W1206 03:59:25.714000 23456244418368 torch/multiprocessing/spawn.py:146] Terminating process 1272007 via signal SIGTERM W1206 03:59:25.715000 23456244418368 torch/multiprocessing/spawn.py:146] Terminating process 1272008 via signal SIGTERM W1206 03:59:25.715000 23456244418368 torch/multiprocessing/spawn.py:146] Terminating process 1272009 via signal SIGTERM W1206 03:59:25.716000 23456244418368 torch/multiprocessing/spawn.py:146] Terminating process 1272011 via signal SIGTERM W1206 03:59:25.716000 23456244418368 torch/multiprocessing/spawn.py:146] Terminating process 1272012 via signal SIGTERM Traceback (most recent call last): File "/mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/runpy.py", line 196, in _run_module_as_main return _run_code(code, main_globals, None, File "/mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/runpy.py", line 86, in _run_code exec(code, run_globals) File "/mnt/home/williamchen/espnet/espnet2/bin/s2t_train.py", line 23, in main() File "/mnt/home/williamchen/espnet/espnet2/bin/s2t_train.py", line 19, in main S2TTask.main(cmd=cmd) File "/mnt/home/williamchen/espnet/espnet2/tasks/abs_task.py", line 1273, in main while not ProcessContext(processes, error_files).join(): File "/mnt/home/williamchen/.local/lib/python3.10/site-packages/torch/multiprocessing/spawn.py", line 178, in join raise ProcessExitedException( torch.multiprocessing.spawn.ProcessExitedException: process 5 terminated with exit code 1 srun: error: cnode7-012: task 0: Exited with exit code 1 [rank15]:[E1206 04:09:16.187204084 ProcessGroupNCCL.cpp:607] [Rank 15] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=408161, OpType=ALLGATHER, NumelIn=1, NumelOut=16, Timeout(ms)=600000) ran for 600003 milliseconds before timing out. [rank8]:[E1206 04:09:16.187207719 ProcessGroupNCCL.cpp:607] [Rank 8] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=408161, OpType=ALLGATHER, NumelIn=1, NumelOut=16, Timeout(ms)=600000) ran for 600003 milliseconds before timing out. [rank9]:[E1206 04:09:16.187224594 ProcessGroupNCCL.cpp:607] [Rank 9] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=408161, OpType=ALLGATHER, NumelIn=1, NumelOut=16, Timeout(ms)=600000) ran for 600000 milliseconds before timing out. [rank14]:[E1206 04:09:16.187808245 ProcessGroupNCCL.cpp:607] [Rank 14] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=408161, OpType=ALLGATHER, NumelIn=1, NumelOut=16, Timeout(ms)=600000) ran for 600004 milliseconds before timing out. [rank13]:[E1206 04:09:16.187210306 ProcessGroupNCCL.cpp:607] [Rank 13] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=408161, OpType=ALLGATHER, NumelIn=1, NumelOut=16, Timeout(ms)=600000) ran for 600004 milliseconds before timing out. [rank12]:[E1206 04:09:16.188964920 ProcessGroupNCCL.cpp:607] [Rank 12] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=408161, OpType=ALLGATHER, NumelIn=1, NumelOut=16, Timeout(ms)=600000) ran for 600006 milliseconds before timing out. [rank10]:[E1206 04:09:16.189312323 ProcessGroupNCCL.cpp:607] [Rank 10] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=408161, OpType=ALLGATHER, NumelIn=1, NumelOut=16, Timeout(ms)=600000) ran for 600005 milliseconds before timing out. [rank14]:[E1206 04:09:16.194322624 ProcessGroupNCCL.cpp:607] [Rank 14] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=408161, OpType=ALLGATHER, NumelIn=1, NumelOut=16, Timeout(ms)=600000) ran for 600011 milliseconds before timing out. [rank11]:[E1206 04:09:16.195818740 ProcessGroupNCCL.cpp:607] [Rank 11] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=408161, OpType=ALLGATHER, NumelIn=1, NumelOut=16, Timeout(ms)=600000) ran for 600008 milliseconds before timing out. [rank14]:[E1206 04:09:16.203083187 ProcessGroupNCCL.cpp:1664] [PG 0 (default_pg) Rank 14] Exception (either an error or timeout) detected by watchdog at work: 408161, last enqueued NCCL work: 408161, last completed NCCL work: 408160. [rank14]:[E1206 04:09:16.203093008 ProcessGroupNCCL.cpp:1709] [PG 0 (default_pg) Rank 14] Timeout at NCCL work: 408161, last enqueued NCCL work: 408161, last completed NCCL work: 408160. [rank14]:[E1206 04:09:16.203097516 ProcessGroupNCCL.cpp:621] [Rank 14] Some NCCL operations have failed or timed out. Due to the asynchronous nature of CUDA kernels, subsequent GPU operations might run on corrupted/incomplete data. [rank8]:[E1206 04:09:16.203998169 ProcessGroupNCCL.cpp:607] [Rank 8] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=408161, OpType=ALLGATHER, NumelIn=1, NumelOut=16, Timeout(ms)=600000) ran for 600022 milliseconds before timing out. [rank8]:[E1206 04:09:16.204078043 ProcessGroupNCCL.cpp:1664] [PG 0 (default_pg) Rank 8] Exception (either an error or timeout) detected by watchdog at work: 408161, last enqueued NCCL work: 408161, last completed NCCL work: 408160. [rank8]:[E1206 04:09:16.204087729 ProcessGroupNCCL.cpp:1709] [PG 0 (default_pg) Rank 8] Timeout at NCCL work: 408161, last enqueued NCCL work: 408161, last completed NCCL work: 408160. [rank8]:[E1206 04:09:16.204093530 ProcessGroupNCCL.cpp:621] [Rank 8] Some NCCL operations have failed or timed out. Due to the asynchronous nature of CUDA kernels, subsequent GPU operations might run on corrupted/incomplete data. [rank13]:[E1206 04:09:16.205115664 ProcessGroupNCCL.cpp:607] [Rank 13] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=408161, OpType=ALLGATHER, NumelIn=1, NumelOut=16, Timeout(ms)=600000) ran for 600023 milliseconds before timing out. [rank13]:[E1206 04:09:16.205187949 ProcessGroupNCCL.cpp:1664] [PG 0 (default_pg) Rank 13] Exception (either an error or timeout) detected by watchdog at work: 408161, last enqueued NCCL work: 408161, last completed NCCL work: 408160. [rank13]:[E1206 04:09:16.205195405 ProcessGroupNCCL.cpp:1709] [PG 0 (default_pg) Rank 13] Timeout at NCCL work: 408161, last enqueued NCCL work: 408161, last completed NCCL work: 408160. [rank13]:[E1206 04:09:16.205199182 ProcessGroupNCCL.cpp:621] [Rank 13] Some NCCL operations have failed or timed out. Due to the asynchronous nature of CUDA kernels, subsequent GPU operations might run on corrupted/incomplete data. [rank8]:[E1206 04:09:16.206196157 ProcessGroupNCCL.cpp:670] [Rank 8] Work WorkNCCL(SeqNum=408161, OpType=ALLGATHER, NumelIn=1, NumelOut=16, Timeout(ms)=600000) timed out in blocking wait (TORCH_NCCL_BLOCKING_WAIT=1). [rank14]:[E1206 04:09:16.206197863 ProcessGroupNCCL.cpp:670] [Rank 14] Work WorkNCCL(SeqNum=408161, OpType=ALLGATHER, NumelIn=1, NumelOut=16, Timeout(ms)=600000) timed out in blocking wait (TORCH_NCCL_BLOCKING_WAIT=1). [rank9]:[E1206 04:09:16.206201054 ProcessGroupNCCL.cpp:670] [Rank 9] Work WorkNCCL(SeqNum=408161, OpType=ALLGATHER, NumelIn=1, NumelOut=16, Timeout(ms)=600000) timed out in blocking wait (TORCH_NCCL_BLOCKING_WAIT=1). [rank13]:[E1206 04:09:16.206203550 ProcessGroupNCCL.cpp:670] [Rank 13] Work WorkNCCL(SeqNum=408161, OpType=ALLGATHER, NumelIn=1, NumelOut=16, Timeout(ms)=600000) timed out in blocking wait (TORCH_NCCL_BLOCKING_WAIT=1). [rank12]:[E1206 04:09:16.206205610 ProcessGroupNCCL.cpp:670] [Rank 12] Work WorkNCCL(SeqNum=408161, OpType=ALLGATHER, NumelIn=1, NumelOut=16, Timeout(ms)=600000) timed out in blocking wait (TORCH_NCCL_BLOCKING_WAIT=1). [rank10]:[E1206 04:09:16.206209324 ProcessGroupNCCL.cpp:670] [Rank 10] Work WorkNCCL(SeqNum=408161, OpType=ALLGATHER, NumelIn=1, NumelOut=16, Timeout(ms)=600000) timed out in blocking wait (TORCH_NCCL_BLOCKING_WAIT=1). [rank11]:[E1206 04:09:16.206212817 ProcessGroupNCCL.cpp:670] [Rank 11] Work WorkNCCL(SeqNum=408161, OpType=ALLGATHER, NumelIn=1, NumelOut=16, Timeout(ms)=600000) timed out in blocking wait (TORCH_NCCL_BLOCKING_WAIT=1). [rank15]:[E1206 04:09:16.206224701 ProcessGroupNCCL.cpp:670] [Rank 15] Work WorkNCCL(SeqNum=408161, OpType=ALLGATHER, NumelIn=1, NumelOut=16, Timeout(ms)=600000) timed out in blocking wait (TORCH_NCCL_BLOCKING_WAIT=1). cnode7-013:736481:746417 [0] NCCL INFO [Service thread] Connection closed by localRank 5 cnode7-013:736481:746417 [0] NCCL INFO [Service thread] Connection closed by localRank 6 cnode7-013:736481:746417 [0] NCCL INFO [Service thread] Connection closed by localRank 7 cnode7-013:736487:746411 [6] NCCL INFO [Service thread] Connection closed by localRank 6 cnode7-013:736487:746411 [6] NCCL INFO [Service thread] Connection closed by localRank 5 cnode7-013:736487:746411 [6] NCCL INFO [Service thread] Connection closed by localRank 7 cnode7-013:736487:746411 [6] NCCL INFO [Service thread] Connection closed by localRank 0 cnode7-013:736486:746409 [5] NCCL INFO [Service thread] Connection closed by localRank 5 cnode7-013:736486:746409 [5] NCCL INFO [Service thread] Connection closed by localRank 6 cnode7-013:736486:746409 [5] NCCL INFO [Service thread] Connection closed by localRank 7 cnode7-013:736486:746409 [5] NCCL INFO [Service thread] Connection closed by localRank 0 cnode7-013:736485:746410 [4] NCCL INFO [Service thread] Connection closed by localRank 5 cnode7-013:736485:746410 [4] NCCL INFO [Service thread] Connection closed by localRank 6 cnode7-013:736485:746410 [4] NCCL INFO [Service thread] Connection closed by localRank 7 cnode7-013:736485:746410 [4] NCCL INFO [Service thread] Connection closed by localRank 0 cnode7-013:736483:746422 [2] NCCL INFO [Service thread] Connection closed by localRank 5 cnode7-013:736483:746422 [2] NCCL INFO [Service thread] Connection closed by localRank 6 cnode7-013:736483:746422 [2] NCCL INFO [Service thread] Connection closed by localRank 7 cnode7-013:736483:746422 [2] NCCL INFO [Service thread] Connection closed by localRank 0 cnode7-013:736482:746416 [1] NCCL INFO [Service thread] Connection closed by localRank 5 cnode7-013:736482:746416 [1] NCCL INFO [Service thread] Connection closed by localRank 6 cnode7-013:736482:746416 [1] NCCL INFO [Service thread] Connection closed by localRank 7 cnode7-013:736482:746416 [1] NCCL INFO [Service thread] Connection closed by localRank 0 cnode7-013:736481:746417 [0] NCCL INFO [Service thread] Connection closed by localRank 0 cnode7-013:736481:746417 [0] NCCL INFO [Service thread] Connection closed by localRank 1 cnode7-013:736484:746415 [3] NCCL INFO [Service thread] Connection closed by localRank 5 cnode7-013:736484:746415 [3] NCCL INFO [Service thread] Connection closed by localRank 6 cnode7-013:736484:746415 [3] NCCL INFO [Service thread] Connection closed by localRank 7 cnode7-013:736484:746415 [3] NCCL INFO [Service thread] Connection closed by localRank 0 cnode7-013:736484:746415 [3] NCCL INFO [Service thread] Connection closed by localRank 1 cnode7-013:736485:746410 [4] NCCL INFO [Service thread] Connection closed by localRank 1 cnode7-013:736482:746416 [1] NCCL INFO [Service thread] Connection closed by localRank 1 cnode7-013:736482:746416 [1] NCCL INFO [Service thread] Connection closed by localRank 3 cnode7-013:736482:746416 [1] NCCL INFO [Service thread] Connection closed by localRank 4 cnode7-013:736483:746422 [2] NCCL INFO [Service thread] Connection closed by localRank 1 cnode7-013:736483:746422 [2] NCCL INFO [Service thread] Connection closed by localRank 3 cnode7-013:736483:746422 [2] NCCL INFO [Service thread] Connection closed by localRank 4 cnode7-013:736484:746415 [3] NCCL INFO [Service thread] Connection closed by localRank 3 cnode7-013:736484:746415 [3] NCCL INFO [Service thread] Connection closed by localRank 4 cnode7-013:736484:746415 [3] NCCL INFO [Service thread] Connection closed by localRank 2 cnode7-013:736488:746408 [7] NCCL INFO [Service thread] Connection closed by localRank 5 cnode7-013:736488:746408 [7] NCCL INFO [Service thread] Connection closed by localRank 6 cnode7-013:736488:746408 [7] NCCL INFO [Service thread] Connection closed by localRank 7 cnode7-013:736488:746408 [7] NCCL INFO [Service thread] Connection closed by localRank 0 cnode7-013:736488:746408 [7] NCCL INFO [Service thread] Connection closed by localRank 1 cnode7-013:736488:746408 [7] NCCL INFO [Service thread] Connection closed by localRank 3 cnode7-013:736488:746408 [7] NCCL INFO [Service thread] Connection closed by localRank 4 cnode7-013:736488:746408 [7] NCCL INFO [Service thread] Connection closed by localRank 2 cnode7-013:736485:746410 [4] NCCL INFO [Service thread] Connection closed by localRank 3 cnode7-013:736485:746410 [4] NCCL INFO [Service thread] Connection closed by localRank 4 cnode7-013:736485:746410 [4] NCCL INFO [Service thread] Connection closed by localRank 2 cnode7-013:736482:746416 [1] NCCL INFO [Service thread] Connection closed by localRank 2 cnode7-013:736481:746417 [0] NCCL INFO [Service thread] Connection closed by localRank 3 cnode7-013:736481:746417 [0] NCCL INFO [Service thread] Connection closed by localRank 4 cnode7-013:736481:746417 [0] NCCL INFO [Service thread] Connection closed by localRank 2 cnode7-013:736487:746411 [6] NCCL INFO [Service thread] Connection closed by localRank 1 cnode7-013:736487:746411 [6] NCCL INFO [Service thread] Connection closed by localRank 3 cnode7-013:736487:746411 [6] NCCL INFO [Service thread] Connection closed by localRank 4 cnode7-013:736487:746411 [6] NCCL INFO [Service thread] Connection closed by localRank 2 cnode7-013:736483:746422 [2] NCCL INFO [Service thread] Connection closed by localRank 2 cnode7-013:736486:746409 [5] NCCL INFO [Service thread] Connection closed by localRank 1 cnode7-013:736486:746409 [5] NCCL INFO [Service thread] Connection closed by localRank 3 cnode7-013:736486:746409 [5] NCCL INFO [Service thread] Connection closed by localRank 4 cnode7-013:736486:746409 [5] NCCL INFO [Service thread] Connection closed by localRank 2 cnode7-013:736487:736487 [6] NCCL INFO comm 0x5555cf9ae150 rank 14 nranks 16 cudaDev 6 busId d1000 - Abort COMPLETE [rank14]:[E1206 04:09:16.574840209 ProcessGroupNCCL.cpp:621] [Rank 14] Some NCCL operations have failed or timed out. Due to the asynchronous nature of CUDA kernels, subsequent GPU operations might run on corrupted/incomplete data. [rank14]:[E1206 04:09:16.574851999 ProcessGroupNCCL.cpp:627] [Rank 14] To avoid data inconsistency, we are taking the entire process down. cnode7-013:736481:736481 [0] NCCL INFO comm 0x5555c59f0f70 rank 8 nranks 16 cudaDev 0 busId 1b000 - Abort COMPLETE [rank8]:[E1206 04:09:16.595897953 ProcessGroupNCCL.cpp:621] [Rank 8] Some NCCL operations have failed or timed out. Due to the asynchronous nature of CUDA kernels, subsequent GPU operations might run on corrupted/incomplete data. [rank8]:[E1206 04:09:16.595910608 ProcessGroupNCCL.cpp:627] [Rank 8] To avoid data inconsistency, we are taking the entire process down. cnode7-013:736484:736484 [3] NCCL INFO comm 0x5555cd20c9e0 rank 11 nranks 16 cudaDev 3 busId 61000 - Abort COMPLETE [rank11]:[E1206 04:09:16.627906490 ProcessGroupNCCL.cpp:621] [Rank 11] Some NCCL operations have failed or timed out. Due to the asynchronous nature of CUDA kernels, subsequent GPU operations might run on corrupted/incomplete data. [rank11]:[E1206 04:09:16.627921314 ProcessGroupNCCL.cpp:627] [Rank 11] To avoid data inconsistency, we are taking the entire process down. [rank11]:[E1206 04:09:16.634898331 ProcessGroupNCCL.cpp:1664] [PG 0 (default_pg) Rank 11] Exception (either an error or timeout) detected by watchdog at work: 408161, last enqueued NCCL work: 408161, last completed NCCL work: 408160. [rank11]:[E1206 04:09:16.634911993 ProcessGroupNCCL.cpp:1709] [PG 0 (default_pg) Rank 11] Timeout at NCCL work: 408161, last enqueued NCCL work: 408161, last completed NCCL work: 408160. [rank11]:[E1206 04:09:16.634916642 ProcessGroupNCCL.cpp:621] [Rank 11] Some NCCL operations have failed or timed out. Due to the asynchronous nature of CUDA kernels, subsequent GPU operations might run on corrupted/incomplete data. cnode7-013:736485:736485 [4] NCCL INFO comm 0x5555cf9ca4a0 rank 12 nranks 16 cudaDev 4 busId 9d000 - Abort COMPLETE [rank12]:[E1206 04:09:16.666723560 ProcessGroupNCCL.cpp:621] [Rank 12] Some NCCL operations have failed or timed out. Due to the asynchronous nature of CUDA kernels, subsequent GPU operations might run on corrupted/incomplete data. [rank12]:[E1206 04:09:16.666739416 ProcessGroupNCCL.cpp:627] [Rank 12] To avoid data inconsistency, we are taking the entire process down. [rank12]:[E1206 04:09:16.668542629 ProcessGroupNCCL.cpp:1664] [PG 0 (default_pg) Rank 12] Exception (either an error or timeout) detected by watchdog at work: 408161, last enqueued NCCL work: 408161, last completed NCCL work: 408160. [rank12]:[E1206 04:09:16.668554846 ProcessGroupNCCL.cpp:1709] [PG 0 (default_pg) Rank 12] Timeout at NCCL work: 408161, last enqueued NCCL work: 408161, last completed NCCL work: 408160. [rank12]:[E1206 04:09:16.668559054 ProcessGroupNCCL.cpp:621] [Rank 12] Some NCCL operations have failed or timed out. Due to the asynchronous nature of CUDA kernels, subsequent GPU operations might run on corrupted/incomplete data. cnode7-013:736486:736486 [5] NCCL INFO comm 0x5555cd22bfa0 rank 13 nranks 16 cudaDev 5 busId c3000 - Abort COMPLETE [rank13]:[E1206 04:09:16.670773994 ProcessGroupNCCL.cpp:621] [Rank 13] Some NCCL operations have failed or timed out. Due to the asynchronous nature of CUDA kernels, subsequent GPU operations might run on corrupted/incomplete data. [rank13]:[E1206 04:09:16.670784224 ProcessGroupNCCL.cpp:627] [Rank 13] To avoid data inconsistency, we are taking the entire process down. cnode7-013:736482:736482 [1] NCCL INFO comm 0x5555cf9ca2c0 rank 9 nranks 16 cudaDev 1 busId 43000 - Abort COMPLETE [rank9]:[E1206 04:09:16.674328398 ProcessGroupNCCL.cpp:621] [Rank 9] Some NCCL operations have failed or timed out. Due to the asynchronous nature of CUDA kernels, subsequent GPU operations might run on corrupted/incomplete data. [rank9]:[E1206 04:09:16.674348921 ProcessGroupNCCL.cpp:627] [Rank 9] To avoid data inconsistency, we are taking the entire process down. cnode7-013:736483:736483 [2] NCCL INFO comm 0x5555cf9ca470 rank 10 nranks 16 cudaDev 2 busId 52000 - Abort COMPLETE [rank10]:[E1206 04:09:16.674831361 ProcessGroupNCCL.cpp:621] [Rank 10] Some NCCL operations have failed or timed out. Due to the asynchronous nature of CUDA kernels, subsequent GPU operations might run on corrupted/incomplete data. [rank10]:[E1206 04:09:16.674847472 ProcessGroupNCCL.cpp:627] [Rank 10] To avoid data inconsistency, we are taking the entire process down. [rank10]:[E1206 04:09:16.676770697 ProcessGroupNCCL.cpp:1664] [PG 0 (default_pg) Rank 10] Exception (either an error or timeout) detected by watchdog at work: 408161, last enqueued NCCL work: 408161, last completed NCCL work: 408160. [rank10]:[E1206 04:09:16.676783155 ProcessGroupNCCL.cpp:1709] [PG 0 (default_pg) Rank 10] Timeout at NCCL work: 408161, last enqueued NCCL work: 408161, last completed NCCL work: 408160. [rank9]:[E1206 04:09:16.676774654 ProcessGroupNCCL.cpp:1664] [PG 0 (default_pg) Rank 9] Exception (either an error or timeout) detected by watchdog at work: 408161, last enqueued NCCL work: 408161, last completed NCCL work: 408160. [rank10]:[E1206 04:09:16.676787760 ProcessGroupNCCL.cpp:621] [Rank 10] Some NCCL operations have failed or timed out. Due to the asynchronous nature of CUDA kernels, subsequent GPU operations might run on corrupted/incomplete data. [rank9]:[E1206 04:09:16.676790237 ProcessGroupNCCL.cpp:1709] [PG 0 (default_pg) Rank 9] Timeout at NCCL work: 408161, last enqueued NCCL work: 408161, last completed NCCL work: 408160. [rank9]:[E1206 04:09:16.676796293 ProcessGroupNCCL.cpp:621] [Rank 9] Some NCCL operations have failed or timed out. Due to the asynchronous nature of CUDA kernels, subsequent GPU operations might run on corrupted/incomplete data. cnode7-013:736488:736488 [7] NCCL INFO comm 0x5555cf9aad40 rank 15 nranks 16 cudaDev 7 busId df000 - Abort COMPLETE [rank15]:[E1206 04:09:16.745397911 ProcessGroupNCCL.cpp:621] [Rank 15] Some NCCL operations have failed or timed out. Due to the asynchronous nature of CUDA kernels, subsequent GPU operations might run on corrupted/incomplete data. [rank15]:[E1206 04:09:16.745417703 ProcessGroupNCCL.cpp:627] [Rank 15] To avoid data inconsistency, we are taking the entire process down. [rank15]:[E1206 04:09:16.747437869 ProcessGroupNCCL.cpp:1664] [PG 0 (default_pg) Rank 15] Exception (either an error or timeout) detected by watchdog at work: 408161, last enqueued NCCL work: 408161, last completed NCCL work: 408160. [rank15]:[E1206 04:09:16.747453766 ProcessGroupNCCL.cpp:1709] [PG 0 (default_pg) Rank 15] Timeout at NCCL work: 408161, last enqueued NCCL work: 408161, last completed NCCL work: 408160. [rank15]:[E1206 04:09:16.747458605 ProcessGroupNCCL.cpp:621] [Rank 15] Some NCCL operations have failed or timed out. Due to the asynchronous nature of CUDA kernels, subsequent GPU operations might run on corrupted/incomplete data. Process SpawnProcess-7: Traceback (most recent call last): File "/mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/multiprocessing/process.py", line 314, in _bootstrap self.run() File "/mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/multiprocessing/process.py", line 108, in run self._target(*self._args, **self._kwargs) File "/mnt/home/williamchen/espnet/espnet2/tasks/abs_task.py", line 1568, in main_worker cls.trainer.run( File "/mnt/home/williamchen/espnet/espnet2/train/deepspeed_trainer.py", line 132, in run cls.train_one_epoch( File "/mnt/home/williamchen/espnet/espnet2/train/deepspeed_trainer.py", line 208, in train_one_epoch stats, weight = recursive_average(stats, weight, True) File "/mnt/home/williamchen/espnet/espnet2/torch_utils/recursive_op.py", line 53, in recursive_average obj = recursive_sum(obj, weight, distributed) File "/mnt/home/williamchen/espnet/espnet2/torch_utils/recursive_op.py", line 14, in recursive_sum return {k: recursive_sum(v, weight, distributed) for k, v in obj.items()} File "/mnt/home/williamchen/espnet/espnet2/torch_utils/recursive_op.py", line 14, in return {k: recursive_sum(v, weight, distributed) for k, v in obj.items()} File "/mnt/home/williamchen/espnet/espnet2/torch_utils/recursive_op.py", line 22, in recursive_sum torch.distributed.all_gather(lst, obj) File "/mnt/home/williamchen/.local/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 79, in wrapper return func(*args, **kwargs) File "/mnt/home/williamchen/.local/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 3113, in all_gather work.wait() torch.distributed.DistBackendError: [Rank 14] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=408161, OpType=ALLGATHER, NumelIn=1, NumelOut=16, Timeout(ms)=600000) ran for 600004 milliseconds before timing out. Process SpawnProcess-2: Traceback (most recent call last): File "/mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/multiprocessing/process.py", line 314, in _bootstrap self.run() File "/mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/multiprocessing/process.py", line 108, in run self._target(*self._args, **self._kwargs) File "/mnt/home/williamchen/espnet/espnet2/tasks/abs_task.py", line 1568, in main_worker cls.trainer.run( File "/mnt/home/williamchen/espnet/espnet2/train/deepspeed_trainer.py", line 132, in run cls.train_one_epoch( File "/mnt/home/williamchen/espnet/espnet2/train/deepspeed_trainer.py", line 208, in train_one_epoch stats, weight = recursive_average(stats, weight, True) File "/mnt/home/williamchen/espnet/espnet2/torch_utils/recursive_op.py", line 53, in recursive_average obj = recursive_sum(obj, weight, distributed) File "/mnt/home/williamchen/espnet/espnet2/torch_utils/recursive_op.py", line 14, in recursive_sum return {k: recursive_sum(v, weight, distributed) for k, v in obj.items()} File "/mnt/home/williamchen/espnet/espnet2/torch_utils/recursive_op.py", line 14, in return {k: recursive_sum(v, weight, distributed) for k, v in obj.items()} File "/mnt/home/williamchen/espnet/espnet2/torch_utils/recursive_op.py", line 22, in recursive_sum torch.distributed.all_gather(lst, obj) File "/mnt/home/williamchen/.local/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 79, in wrapper return func(*args, **kwargs) File "/mnt/home/williamchen/.local/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 3113, in all_gather work.wait() torch.distributed.DistBackendError: [Rank 9] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=408161, OpType=ALLGATHER, NumelIn=1, NumelOut=16, Timeout(ms)=600000) ran for 600000 milliseconds before timing out. Process SpawnProcess-3: Traceback (most recent call last): File "/mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/multiprocessing/process.py", line 314, in _bootstrap self.run() File "/mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/multiprocessing/process.py", line 108, in run self._target(*self._args, **self._kwargs) File "/mnt/home/williamchen/espnet/espnet2/tasks/abs_task.py", line 1568, in main_worker cls.trainer.run( File "/mnt/home/williamchen/espnet/espnet2/train/deepspeed_trainer.py", line 132, in run cls.train_one_epoch( File "/mnt/home/williamchen/espnet/espnet2/train/deepspeed_trainer.py", line 208, in train_one_epoch stats, weight = recursive_average(stats, weight, True) File "/mnt/home/williamchen/espnet/espnet2/torch_utils/recursive_op.py", line 53, in recursive_average obj = recursive_sum(obj, weight, distributed) File "/mnt/home/williamchen/espnet/espnet2/torch_utils/recursive_op.py", line 14, in recursive_sum return {k: recursive_sum(v, weight, distributed) for k, v in obj.items()} File "/mnt/home/williamchen/espnet/espnet2/torch_utils/recursive_op.py", line 14, in return {k: recursive_sum(v, weight, distributed) for k, v in obj.items()} File "/mnt/home/williamchen/espnet/espnet2/torch_utils/recursive_op.py", line 22, in recursive_sum torch.distributed.all_gather(lst, obj) File "/mnt/home/williamchen/.local/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 79, in wrapper return func(*args, **kwargs) File "/mnt/home/williamchen/.local/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 3113, in all_gather work.wait() torch.distributed.DistBackendError: [Rank 10] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=408161, OpType=ALLGATHER, NumelIn=1, NumelOut=16, Timeout(ms)=600000) ran for 600005 milliseconds before timing out. Process SpawnProcess-1: Traceback (most recent call last): File "/mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/multiprocessing/process.py", line 314, in _bootstrap self.run() File "/mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/multiprocessing/process.py", line 108, in run self._target(*self._args, **self._kwargs) File "/mnt/home/williamchen/espnet/espnet2/tasks/abs_task.py", line 1568, in main_worker cls.trainer.run( File "/mnt/home/williamchen/espnet/espnet2/train/deepspeed_trainer.py", line 132, in run cls.train_one_epoch( File "/mnt/home/williamchen/espnet/espnet2/train/deepspeed_trainer.py", line 208, in train_one_epoch stats, weight = recursive_average(stats, weight, True) File "/mnt/home/williamchen/espnet/espnet2/torch_utils/recursive_op.py", line 53, in recursive_average obj = recursive_sum(obj, weight, distributed) File "/mnt/home/williamchen/espnet/espnet2/torch_utils/recursive_op.py", line 14, in recursive_sum return {k: recursive_sum(v, weight, distributed) for k, v in obj.items()} File "/mnt/home/williamchen/espnet/espnet2/torch_utils/recursive_op.py", line 14, in return {k: recursive_sum(v, weight, distributed) for k, v in obj.items()} File "/mnt/home/williamchen/espnet/espnet2/torch_utils/recursive_op.py", line 22, in recursive_sum torch.distributed.all_gather(lst, obj) File "/mnt/home/williamchen/.local/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 79, in wrapper return func(*args, **kwargs) File "/mnt/home/williamchen/.local/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 3113, in all_gather work.wait() torch.distributed.DistBackendError: [Rank 8] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=408161, OpType=ALLGATHER, NumelIn=1, NumelOut=16, Timeout(ms)=600000) ran for 600003 milliseconds before timing out. Process SpawnProcess-6: Traceback (most recent call last): File "/mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/multiprocessing/process.py", line 314, in _bootstrap self.run() File "/mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/multiprocessing/process.py", line 108, in run self._target(*self._args, **self._kwargs) File "/mnt/home/williamchen/espnet/espnet2/tasks/abs_task.py", line 1568, in main_worker cls.trainer.run( File "/mnt/home/williamchen/espnet/espnet2/train/deepspeed_trainer.py", line 132, in run cls.train_one_epoch( File "/mnt/home/williamchen/espnet/espnet2/train/deepspeed_trainer.py", line 208, in train_one_epoch stats, weight = recursive_average(stats, weight, True) File "/mnt/home/williamchen/espnet/espnet2/torch_utils/recursive_op.py", line 53, in recursive_average obj = recursive_sum(obj, weight, distributed) File "/mnt/home/williamchen/espnet/espnet2/torch_utils/recursive_op.py", line 14, in recursive_sum return {k: recursive_sum(v, weight, distributed) for k, v in obj.items()} File "/mnt/home/williamchen/espnet/espnet2/torch_utils/recursive_op.py", line 14, in return {k: recursive_sum(v, weight, distributed) for k, v in obj.items()} File "/mnt/home/williamchen/espnet/espnet2/torch_utils/recursive_op.py", line 22, in recursive_sum torch.distributed.all_gather(lst, obj) File "/mnt/home/williamchen/.local/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 79, in wrapper return func(*args, **kwargs) File "/mnt/home/williamchen/.local/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 3113, in all_gather work.wait() torch.distributed.DistBackendError: [Rank 13] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=408161, OpType=ALLGATHER, NumelIn=1, NumelOut=16, Timeout(ms)=600000) ran for 600004 milliseconds before timing out. Process SpawnProcess-5: Traceback (most recent call last): File "/mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/multiprocessing/process.py", line 314, in _bootstrap self.run() File "/mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/multiprocessing/process.py", line 108, in run self._target(*self._args, **self._kwargs) File "/mnt/home/williamchen/espnet/espnet2/tasks/abs_task.py", line 1568, in main_worker cls.trainer.run( File "/mnt/home/williamchen/espnet/espnet2/train/deepspeed_trainer.py", line 132, in run cls.train_one_epoch( File "/mnt/home/williamchen/espnet/espnet2/train/deepspeed_trainer.py", line 208, in train_one_epoch stats, weight = recursive_average(stats, weight, True) File "/mnt/home/williamchen/espnet/espnet2/torch_utils/recursive_op.py", line 53, in recursive_average obj = recursive_sum(obj, weight, distributed) File "/mnt/home/williamchen/espnet/espnet2/torch_utils/recursive_op.py", line 14, in recursive_sum return {k: recursive_sum(v, weight, distributed) for k, v in obj.items()} File "/mnt/home/williamchen/espnet/espnet2/torch_utils/recursive_op.py", line 14, in return {k: recursive_sum(v, weight, distributed) for k, v in obj.items()} File "/mnt/home/williamchen/espnet/espnet2/torch_utils/recursive_op.py", line 22, in recursive_sum torch.distributed.all_gather(lst, obj) File "/mnt/home/williamchen/.local/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 79, in wrapper return func(*args, **kwargs) File "/mnt/home/williamchen/.local/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 3113, in all_gather work.wait() torch.distributed.DistBackendError: [Rank 12] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=408161, OpType=ALLGATHER, NumelIn=1, NumelOut=16, Timeout(ms)=600000) ran for 600006 milliseconds before timing out. Process SpawnProcess-8: Traceback (most recent call last): File "/mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/multiprocessing/process.py", line 314, in _bootstrap self.run() File "/mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/multiprocessing/process.py", line 108, in run self._target(*self._args, **self._kwargs) File "/mnt/home/williamchen/espnet/espnet2/tasks/abs_task.py", line 1568, in main_worker cls.trainer.run( File "/mnt/home/williamchen/espnet/espnet2/train/deepspeed_trainer.py", line 132, in run cls.train_one_epoch( File "/mnt/home/williamchen/espnet/espnet2/train/deepspeed_trainer.py", line 208, in train_one_epoch stats, weight = recursive_average(stats, weight, True) File "/mnt/home/williamchen/espnet/espnet2/torch_utils/recursive_op.py", line 53, in recursive_average obj = recursive_sum(obj, weight, distributed) File "/mnt/home/williamchen/espnet/espnet2/torch_utils/recursive_op.py", line 14, in recursive_sum return {k: recursive_sum(v, weight, distributed) for k, v in obj.items()} File "/mnt/home/williamchen/espnet/espnet2/torch_utils/recursive_op.py", line 14, in return {k: recursive_sum(v, weight, distributed) for k, v in obj.items()} File "/mnt/home/williamchen/espnet/espnet2/torch_utils/recursive_op.py", line 22, in recursive_sum torch.distributed.all_gather(lst, obj) File "/mnt/home/williamchen/.local/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 79, in wrapper return func(*args, **kwargs) File "/mnt/home/williamchen/.local/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 3113, in all_gather work.wait() torch.distributed.DistBackendError: [Rank 15] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=408161, OpType=ALLGATHER, NumelIn=1, NumelOut=16, Timeout(ms)=600000) ran for 600003 milliseconds before timing out. Process SpawnProcess-4: Traceback (most recent call last): File "/mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/multiprocessing/process.py", line 314, in _bootstrap self.run() File "/mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/multiprocessing/process.py", line 108, in run self._target(*self._args, **self._kwargs) File "/mnt/home/williamchen/espnet/espnet2/tasks/abs_task.py", line 1568, in main_worker cls.trainer.run( File "/mnt/home/williamchen/espnet/espnet2/train/deepspeed_trainer.py", line 132, in run cls.train_one_epoch( File "/mnt/home/williamchen/espnet/espnet2/train/deepspeed_trainer.py", line 208, in train_one_epoch stats, weight = recursive_average(stats, weight, True) File "/mnt/home/williamchen/espnet/espnet2/torch_utils/recursive_op.py", line 53, in recursive_average obj = recursive_sum(obj, weight, distributed) File "/mnt/home/williamchen/espnet/espnet2/torch_utils/recursive_op.py", line 14, in recursive_sum return {k: recursive_sum(v, weight, distributed) for k, v in obj.items()} File "/mnt/home/williamchen/espnet/espnet2/torch_utils/recursive_op.py", line 14, in return {k: recursive_sum(v, weight, distributed) for k, v in obj.items()} File "/mnt/home/williamchen/espnet/espnet2/torch_utils/recursive_op.py", line 22, in recursive_sum torch.distributed.all_gather(lst, obj) File "/mnt/home/williamchen/.local/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 79, in wrapper return func(*args, **kwargs) File "/mnt/home/williamchen/.local/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 3113, in all_gather work.wait() torch.distributed.DistBackendError: [Rank 11] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=408161, OpType=ALLGATHER, NumelIn=1, NumelOut=16, Timeout(ms)=600000) ran for 600008 milliseconds before timing out. W1206 04:09:27.023000 23456244418368 torch/multiprocessing/spawn.py:146] Terminating process 736482 via signal SIGTERM W1206 04:09:27.024000 23456244418368 torch/multiprocessing/spawn.py:146] Terminating process 736484 via signal SIGTERM W1206 04:09:27.024000 23456244418368 torch/multiprocessing/spawn.py:146] Terminating process 736485 via signal SIGTERM W1206 04:09:27.024000 23456244418368 torch/multiprocessing/spawn.py:146] Terminating process 736487 via signal SIGTERM Traceback (most recent call last): File "/mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/runpy.py", line 196, in _run_module_as_main return _run_code(code, main_globals, None, File "/mnt/home/williamchen/espnet/tools/miniconda/envs/espnet/lib/python3.10/runpy.py", line 86, in _run_code exec(code, run_globals) File "/mnt/home/williamchen/espnet/espnet2/bin/s2t_train.py", line 23, in main() File "/mnt/home/williamchen/espnet/espnet2/bin/s2t_train.py", line 19, in main S2TTask.main(cmd=cmd) File "/mnt/home/williamchen/espnet/espnet2/tasks/abs_task.py", line 1273, in main while not ProcessContext(processes, error_files).join(): File "/mnt/home/williamchen/.local/lib/python3.10/site-packages/torch/multiprocessing/spawn.py", line 178, in join raise ProcessExitedException( torch.multiprocessing.spawn.ProcessExitedException: process 5 terminated with exit code 1 srun: error: cnode7-013: task 1: Exited with exit code 1 # Accounting: begin_time=1733387579 # Accounting: end_time=1733429369 # Accounting: time=41790 threads=1 # Finished at Fri Dec 6 04:09:29 CST 2024 with status 1