KoReason/Qwen3-ot3-en-sampled

See axolotl config

axolotl version: 0.11.0

base_model: Qwen/Qwen3-4B


load_in_8bit: false
load_in_4bit: false

datasets:
  - path: train_ot3_ko_sampled.jsonl
    type: chat_template

dataset_prepared_path: preprocess
val_set_size: 0.01
output_dir: ./outputs
dataloader_num_workers: 24

adapter: 
lora_model_dir:

sequence_len: 16384
sample_packing: false
eval_sample_packing: false
pad_to_sequence_len: false

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

wandb_project: 
wandb_entity:
wandb_watch:
wandb_name: 
wandb_log_model:
hub_model_id: 

gradient_accumulation_steps: 8
micro_batch_size: 8
num_epochs: 5
optimizer: adamw_torch_fused
lr_scheduler: cosine
learning_rate: 4e-5

bf16: auto
tf32: false

gradient_checkpointing:
resume_from_checkpoint:
logging_steps: 5
flash_attention: true

warmup_ratio: 0.05
weight_decay: 0.01
evals_per_epoch: 0
saves_per_epoch: 1

fsdp:
  - full_shard
  - auto_wrap
fsdp_config:
  fsdp_state_dict_type: FULL_STATE_DICT
  fsdp_transformer_layer_cls_to_wrap: Qwen3DecoderLayer
  fsdp_activation_checkpointing: true

outputs

This model is a fine-tuned version of Qwen/Qwen3-4B on the train_ot3_ko_sampled.jsonl dataset.

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 4e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• gradient_accumulation_steps: 8
• total_train_batch_size: 512
• total_eval_batch_size: 64
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 55
• training_steps: 1106

Training results

Framework versions

• Transformers 4.53.1
• Pytorch 2.5.1
• Datasets 4.0.0
• Tokenizers 0.21.2