| bf16: true | |
| cutoff_len: 1024 | |
| dataset: XeTute/SStory-Gen-EN_ZH,MatanP/emotion_mapped_story_dataset,webnovel,jaydenccc/AI_Storyteller_Dataset | |
| dataset_dir: data | |
| ddp_timeout: 180000000 | |
| do_train: true | |
| finetuning_type: full | |
| flash_attn: auto | |
| gradient_accumulation_steps: 1 | |
| include_num_input_tokens_seen: true | |
| learning_rate: 1.0e-06 | |
| logging_steps: 100 | |
| lr_scheduler_type: cosine | |
| max_grad_norm: 1.0 | |
| max_samples: 1000000000 | |
| model_name_or_path: XeTute/Phantasor_V0.2-137M | |
| num_train_epochs: 12.0 | |
| optim: sgd | |
| output_dir: saves\GPT-2-Small\full\10-02-2025 | |
| packing: false | |
| per_device_train_batch_size: 1 | |
| plot_loss: true | |
| preprocessing_num_workers: 16 | |
| report_to: none | |
| save_steps: 5000 | |
| stage: sft | |
| template: alpaca | |
| trust_remote_code: true | |
| warmup_steps: 10 | |