| { |
| "model_name_or_path": "/inspire/hdd/ws-f4d69b29-e0a5-44e6-bd92-acf4de9990f0/public-project/jiangchangyue-240114020171/workspace/jcy/reasoning_safety/Thought_Aligner/warmup_model/warmup-qwen-7b", |
| "max_length": 2048, |
| "trust_remote_code": true, |
| "train_datasets": [ |
| [ |
| "correction-json", |
| { |
| "proportion": 1.0, |
| "path": "/inspire/hdd/ws-f4d69b29-e0a5-44e6-bd92-acf4de9990f0/public-project/jiangchangyue-240114020171/workspace/jcy/reasoning_safety/data/train_data.json" |
| } |
| ] |
| ], |
| "eval_datasets": null, |
| "epochs": 3, |
| "per_device_train_batch_size": 4, |
| "per_device_eval_batch_size": 4, |
| "gradient_accumulation_steps": 8, |
| "gradient_checkpointing": true, |
| "lr": 2e-05, |
| "lr_scheduler_type": "cosine", |
| "lr_warmup_ratio": 0.03, |
| "weight_decay": 0.0, |
| "seed": 42, |
| "fp16": false, |
| "bf16": true, |
| "tf32": true, |
| "eval_strategy": "epoch", |
| "eval_interval": 1000000, |
| "need_eval": false, |
| "eval_split_ratio": null, |
| "output_dir": "/inspire/hdd/ws-f4d69b29-e0a5-44e6-bd92-acf4de9990f0/public-project/jiangchangyue-240114020171/workspace/jcy/reasoning_safety/Thought_Aligner/sft_model/thought-aligner-qwen-7b", |
| "log_type": "wandb", |
| "log_dir": "/inspire/hdd/ws-f4d69b29-e0a5-44e6-bd92-acf4de9990f0/public-project/jiangchangyue-240114020171/workspace/jcy/reasoning_safety/Thought_Aligner/sft_model/thought-aligner-qwen-7b", |
| "log_project": "Aligner-SFT", |
| "log_run_name": "sft-2025-03-10-22-03-10", |
| "save_16bit": true, |
| "save_interval": 1000000, |
| "local_rank": 0, |
| "zero_stage": 3, |
| "offload": "none", |
| "deepspeed": false, |
| "deepspeed_config": null, |
| "deepscale": false, |
| "deepscale_config": null, |
| "global_rank": 0, |
| "device": { |
| "type": "torch.device", |
| "repr": "device(type='cuda', index=0)" |
| }, |
| "num_update_steps_per_epoch": 86, |
| "total_training_steps": 258 |
| } |
