| ### model | |
| model_name_or_path: /maindata/data/shared/public/aigame/nlp_models/Qwen2.5-Coder-7B-Instruct | |
| ### method | |
| stage: sft | |
| do_train: true | |
| finetuning_type: full | |
| deepspeed: examples/deepspeed/ds_z3_config.json | |
| ### dataset | |
| dataset: re_arc | |
| template: qwen | |
| cutoff_len: 8192 | |
| max_samples: 1000000000 | |
| overwrite_cache: true | |
| preprocessing_num_workers: 16 | |
| ### output | |
| output_dir: saves/qwen2.5-coder-7b/full/sft | |
| logging_steps: 5 | |
| save_steps: 0.25 | |
| plot_loss: true | |
| overwrite_output_dir: true | |
| ### train | |
| per_device_train_batch_size: 2 | |
| gradient_accumulation_steps: 2 | |
| learning_rate: 1.0e-5 | |
| num_train_epochs: 4.0 | |
| lr_scheduler_type: cosine | |
| warmup_ratio: 0.1 | |
| bf16: true | |
| ddp_timeout: 180000000 | |
| ### eval | |
| val_size: 0.1 | |
| per_device_eval_batch_size: 2 | |
| eval_strategy: steps | |
| eval_steps: 50 | |