| { | |
| "attention_dropout": 0.0, | |
| "base_model_name_or_path": "Qwen/Qwen-7B-Chat", | |
| "calibration_type": "transformer", | |
| "feature_key": "hidden_states", | |
| "freeze_base_model": true, | |
| "hidden_act": "silu", | |
| "in_features": 4096, | |
| "inference_mode": true, | |
| "init_temperature": 1.0, | |
| "intermediate_size": 11008, | |
| "label_smoothing": 1.0, | |
| "label_smoothing_type": "uniform", | |
| "layer_idx": 33, | |
| "log_auxiliary_info": true, | |
| "loss_type": "selective_smoothing", | |
| "max_position_embeddings": 4096, | |
| "normalize_logits": false, | |
| "num_attention_heads": 32, | |
| "num_key_value_heads": 32, | |
| "smooth_loss_weight": 0.3, | |
| "smoothing_topk": 5, | |
| "task_type": "CAUSAL_LM" | |
| } |