| { | |
| "model_name_or_path": "models/Qwen3.5-0.8B", | |
| "revision": "2fc06364715b967f1860aea9cf38778875588b17", | |
| "dtype": "bfloat16", | |
| "attn_implementation": "sdpa", | |
| "workspace_tokens": 128, | |
| "loops": 3, | |
| "loop_scope": "middle_workspace", | |
| "loop_start_block": 1, | |
| "loop_end_block": 24, | |
| "detach_loop_state": false, | |
| "classification_loss_scope": "all_loops", | |
| "aux_weight": 0.3, | |
| "aux_dim": 256, | |
| "aux_heads": 8, | |
| "aux_max_tokens": 4096, | |
| "aux_chunk_size": 1024, | |
| "aux_response_chunk_size": 16, | |
| "freeze_vision": false, | |
| "gradient_checkpointing": true, | |
| "training_stage": 1, | |
| "stage1_objective": "cot_only" | |
| } | |