s300_shuff100 / evalonlyhindi_indicwav2vec_MUCS_warmup500_s300shuff100_2142383.out
Priyanship's picture
End of training
c75a43d verified
wandb: Currently logged in as: priyanshi-pal (priyanshipal). Use `wandb login --relogin` to force relogin
wandb: wandb version 0.17.7 is available! To upgrade, please run:
wandb: $ pip install wandb --upgrade
wandb: Tracking run with wandb version 0.17.6
wandb: Run data is saved locally in /scratch/elec/t405-puhe/p/palp3/MUCS/wandb/run-20240822_150154-1kodfy70
wandb: Run `wandb offline` to turn off syncing.
wandb: Syncing run eval_pd2000_s300_shuff100_hindi
wandb: ⭐️ View project at https://wandb.ai/priyanshipal/huggingface
wandb: 🚀 View run at https://wandb.ai/priyanshipal/huggingface/runs/1kodfy70
/scratch/work/palp3/myenv/lib/python3.11/site-packages/transformers/training_args.py:1525: FutureWarning: `evaluation_strategy` is deprecated and will be removed in version 4.46 of 🤗 Transformers. Use `eval_strategy` instead
warnings.warn(
/scratch/work/palp3/myenv/lib/python3.11/site-packages/transformers/models/auto/configuration_auto.py:957: FutureWarning: The `use_auth_token` argument is deprecated and will be removed in v5 of Transformers. Please use `token` instead.
warnings.warn(
/scratch/work/palp3/myenv/lib/python3.11/site-packages/transformers/models/auto/feature_extraction_auto.py:329: FutureWarning: The `use_auth_token` argument is deprecated and will be removed in v5 of Transformers. Please use `token` instead.
warnings.warn(
/scratch/work/palp3/myenv/lib/python3.11/site-packages/accelerate/accelerator.py:488: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead.
self.scaler = torch.cuda.amp.GradScaler(**kwargs)
max_steps is given, it will override any value given in num_train_epochs
Wav2Vec2CTCTokenizer(name_or_path='', vocab_size=149, model_max_length=1000000000000000019884624838656, is_fast=False, padding_side='right', truncation_side='right', special_tokens={'bos_token': '<s>', 'eos_token': '</s>', 'unk_token': '[UNK]', 'pad_token': '[PAD]'}, clean_up_tokenization_spaces=True), added_tokens_decoder={
147: AddedToken("[UNK]", rstrip=True, lstrip=True, single_word=False, normalized=False, special=False),
148: AddedToken("[PAD]", rstrip=True, lstrip=True, single_word=False, normalized=False, special=False),
149: AddedToken("<s>", rstrip=False, lstrip=False, single_word=False, normalized=False, special=True),
150: AddedToken("</s>", rstrip=False, lstrip=False, single_word=False, normalized=False, special=True),
}
CHECK MODEL PARAMS Wav2Vec2ForCTC(
(wav2vec2): Wav2Vec2Model(
(feature_extractor): Wav2Vec2FeatureEncoder(
(conv_layers): ModuleList(
(0): Wav2Vec2LayerNormConvLayer(
(conv): Conv1d(1, 512, kernel_size=(10,), stride=(5,))
(layer_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True)
(activation): GELUActivation()
)
(1-4): 4 x Wav2Vec2LayerNormConvLayer(
(conv): Conv1d(512, 512, kernel_size=(3,), stride=(2,))
(layer_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True)
(activation): GELUActivation()
)
(5-6): 2 x Wav2Vec2LayerNormConvLayer(
(conv): Conv1d(512, 512, kernel_size=(2,), stride=(2,))
(layer_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True)
(activation): GELUActivation()
)
)
)
(feature_projection): Wav2Vec2FeatureProjection(
(layer_norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True)
(projection): Linear(in_features=512, out_features=1024, bias=True)
(dropout): Dropout(p=0.0, inplace=False)
)
(encoder): Wav2Vec2EncoderStableLayerNorm(
(pos_conv_embed): Wav2Vec2PositionalConvEmbedding(
(conv): ParametrizedConv1d(
1024, 1024, kernel_size=(128,), stride=(1,), padding=(64,), groups=16
(parametrizations): ModuleDict(
(weight): ParametrizationList(
(0): _WeightNorm()
)
)
)
(padding): Wav2Vec2SamePadLayer()
(activation): GELUActivation()
)
(layer_norm): LayerNorm((1024,), eps=1e-05, elementwise_affine=True)
(dropout): Dropout(p=0.0, inplace=False)
(layers): ModuleList(
(0-23): 24 x Wav2Vec2EncoderLayerStableLayerNorm(
(attention): Wav2Vec2SdpaAttention(
(k_proj): Linear(in_features=1024, out_features=1024, bias=True)
(v_proj): Linear(in_features=1024, out_features=1024, bias=True)
(q_proj): Linear(in_features=1024, out_features=1024, bias=True)
(out_proj): Linear(in_features=1024, out_features=1024, bias=True)
)
(dropout): Dropout(p=0.0, inplace=False)
(layer_norm): LayerNorm((1024,), eps=1e-05, elementwise_affine=True)
(feed_forward): Wav2Vec2FeedForward(
(intermediate_dropout): Dropout(p=0.0, inplace=False)
(intermediate_dense): Linear(in_features=1024, out_features=4096, bias=True)
(intermediate_act_fn): GELUActivation()
(output_dense): Linear(in_features=4096, out_features=1024, bias=True)
(output_dropout): Dropout(p=0.0, inplace=False)
)
(final_layer_norm): LayerNorm((1024,), eps=1e-05, elementwise_affine=True)
)
)
)
)
(dropout): Dropout(p=0.0, inplace=False)
(lm_head): Linear(in_features=1024, out_features=151, bias=True)
)
08/22/2024 15:02:06 - INFO - __main__ - *** Evaluate ***
/scratch/work/palp3/myenv/lib/python3.11/site-packages/transformers/models/wav2vec2/processing_wav2vec2.py:157: UserWarning: `as_target_processor` is deprecated and will be removed in v5 of Transformers. You can process your labels by using the argument `text` of the regular `__call__` method (either in the same call as your audio inputs, or in a separate call.
warnings.warn(
0%| | 0/36 [00:00<?, ?it/s] 6%|▌ | 2/36 [00:01<00:26, 1.27it/s] 8%|▊ | 3/36 [00:02<00:33, 1.02s/it] 11%|█ | 4/36 [00:04<00:41, 1.30s/it] 14%|█▍ | 5/36 [00:06<00:43, 1.40s/it] 17%|█▋ | 6/36 [00:07<00:41, 1.39s/it] 19%|█▉ | 7/36 [00:08<00:35, 1.22s/it] 22%|██▏ | 8/36 [00:09<00:27, 1.01it/s] 25%|██▌ | 9/36 [00:09<00:23, 1.17it/s] 28%|██▊ | 10/36 [00:10<00:21, 1.22it/s] 31%|███ | 11/36 [00:11<00:20, 1.20it/s] 33%|███▎ | 12/36 [00:11<00:19, 1.21it/s] 36%|███▌ | 13/36 [00:12<00:18, 1.23it/s] 39%|███▉ | 14/36 [00:13<00:15, 1.43it/s] 42%|████▏ | 15/36 [00:13<00:12, 1.63it/s] 44%|████▍ | 16/36 [00:14<00:11, 1.79it/s] 47%|████▋ | 17/36 [00:14<00:10, 1.89it/s] 50%|█████ | 18/36 [00:15<00:10, 1.79it/s] 53%|█████▎ | 19/36 [00:15<00:09, 1.74it/s] 56%|█████▌ | 20/36 [00:16<00:08, 1.79it/s] 58%|█████▊ | 21/36 [00:16<00:07, 1.94it/s] 61%|██████ | 22/36 [00:17<00:07, 1.94it/s] 64%|██████▍ | 23/36 [00:17<00:06, 1.91it/s] 67%|██████▋ | 24/36 [00:18<00:06, 1.88it/s] 69%|██████▉ | 25/36 [00:18<00:06, 1.81it/s] 72%|███████▏ | 26/36 [00:19<00:05, 1.84it/s] 75%|███████▌ | 27/36 [00:19<00:04, 1.94it/s] 78%|███████▊ | 28/36 [00:20<00:05, 1.59it/s] 81%|████████ | 29/36 [00:22<00:06, 1.08it/s] 83%|████████▎ | 30/36 [00:23<00:06, 1.08s/it] 86%|████████▌ | 31/36 [00:25<00:06, 1.35s/it] 89%|████████▉ | 32/36 [00:26<00:04, 1.13s/it] 92%|█████████▏| 33/36 [00:27<00:02, 1.03it/s] 94%|█████████▍| 34/36 [00:27<00:01, 1.20it/s] 97%|█████████▋| 35/36 [00:28<00:00, 1.35it/s] 100%|██████████| 36/36 [00:28<00:00, 1.65it/s] 100%|██████████| 36/36 [00:30<00:00, 1.20it/s]
Printing predictions for a few samples:
Sample 1:
Reference: हम उनका उपयोग ऐसे ही कर सकते हैं या आवश्यकता अनुसार कुछ बदलाव करके उपयोग कर सकते हैं
######
Prediction:
Sample 2:
Reference: अतः शीर्षक इस तरह से जोड़ सकते हैं
######
Prediction:
Sample 3:
Reference: प्रेसेंटेशन के अंत में आपने स्लाइड की एक कॉपी बना ली है
######
Prediction:
Sample 4:
Reference: चलिए अब फोंट्स और फोंट्स को फॉर्मेट करने के कुछ तरीके देखते हैं
######
Prediction:
Sample 5:
Reference: यह एक डायलॉग बॉक्स खोलेगा जिसमें हम अपनी आवश्यकतानुसार फॉन्ट स्टाइल और साइज़ सेट कर सकते हैं
######
Prediction:
last Reference string यह स्क्रिप्ट लता द्वारा अनुवादित है आईआईटी मुंबई की ओर से मैं रवि कुमार अब आपसे विदा लेता हूँहमसे जुड़ने के लिए धन्यवाद
last prediction string
***** eval metrics *****
eval_cer = 1.0
eval_loss = nan
eval_model_preparation_time = 0.0046
eval_runtime = 0:00:39.88
eval_samples = 572
eval_samples_per_second = 14.34
eval_steps_per_second = 0.902
eval_wer = 1.0
training_args.bin: 0%| | 0.00/5.43k [00:00<?, ?B/s] training_args.bin: 100%|██████████| 5.43k/5.43k [00:00<00:00, 28.4kB/s]
wandb: - 0.005 MB of 0.005 MB uploaded wandb: \ 0.036 MB of 0.036 MB uploaded wandb:
wandb: Run history:
wandb: eval/cer ▁
wandb: eval/model_preparation_time ▁
wandb: eval/runtime ▁
wandb: eval/samples_per_second ▁
wandb: eval/steps_per_second ▁
wandb: eval/wer ▁
wandb: eval_cer ▁
wandb: eval_model_preparation_time ▁
wandb: eval_runtime ▁
wandb: eval_samples ▁
wandb: eval_samples_per_second ▁
wandb: eval_steps_per_second ▁
wandb: eval_wer ▁
wandb: train/global_step ▁▁
wandb:
wandb: Run summary:
wandb: eval/cer 1.0
wandb: eval/loss nan
wandb: eval/model_preparation_time 0.0046
wandb: eval/runtime 39.8895
wandb: eval/samples_per_second 14.34
wandb: eval/steps_per_second 0.902
wandb: eval/wer 1.0
wandb: eval_cer 1.0
wandb: eval_loss nan
wandb: eval_model_preparation_time 0.0046
wandb: eval_runtime 39.8895
wandb: eval_samples 572
wandb: eval_samples_per_second 14.34
wandb: eval_steps_per_second 0.902
wandb: eval_wer 1.0
wandb: train/global_step 0
wandb:
wandb: 🚀 View run eval_pd2000_s300_shuff100_hindi at: https://wandb.ai/priyanshipal/huggingface/runs/1kodfy70
wandb: ⭐️ View project at: https://wandb.ai/priyanshipal/huggingface
wandb: Synced 6 W&B file(s), 0 media file(s), 0 artifact file(s) and 0 other file(s)
wandb: Find logs at: ./wandb/run-20240822_150154-1kodfy70/logs
wandb: WARNING The new W&B backend becomes opt-out in version 0.18.0; try it out with `wandb.require("core")`! See https://wandb.me/wandb-core for more information.