| { | |
| "_name_or_path": "/checkpoint/mengzhouxia/space2/out/squad/layerdistillv3_prunehidden/squad_l0_headint_nosvd_layerpuning_layerdistillv3_prunehidden_seed57_distilltemp2_pretrain6000_warpup12000_ts0.6_cealpha0.5_20epochs/best", | |
| "architectures": [ | |
| "NewBertForQuestionAnswering" | |
| ], | |
| "attention_probs_dropout_prob": 0.1, | |
| "decompose_qk": false, | |
| "decompose_vo": false, | |
| "do_distill": true, | |
| "do_emb_distill": false, | |
| "do_layer_distill": true, | |
| "do_mha_distill": false, | |
| "do_mha_layer_distill": false, | |
| "gradient_checkpointing": false, | |
| "hidden_act": "gelu", | |
| "hidden_dropout_prob": 0.1, | |
| "hidden_size": 768, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 3072, | |
| "layer_norm_eps": 1e-12, | |
| "max_position_embeddings": 512, | |
| "model_type": "bert", | |
| "num_attention_heads": 12, | |
| "num_hidden_layers": 12, | |
| "output_attentions": true, | |
| "output_hidden_states": true, | |
| "pad_token_id": 0, | |
| "position_embedding_type": "absolute", | |
| "pruned_heads": { | |
| "0": [ | |
| 2, | |
| 5 | |
| ], | |
| "1": [ | |
| 0, | |
| 2, | |
| 3 | |
| ], | |
| "2": [ | |
| 8, | |
| 11, | |
| 4, | |
| 7 | |
| ], | |
| "3": [ | |
| 2, | |
| 4 | |
| ], | |
| "4": [ | |
| 8 | |
| ], | |
| "5": [ | |
| 1, | |
| 2, | |
| 11 | |
| ], | |
| "6": [ | |
| 2, | |
| 3 | |
| ], | |
| "7": [ | |
| 11, | |
| 3, | |
| 6, | |
| 7 | |
| ], | |
| "8": [ | |
| 0, | |
| 4 | |
| ], | |
| "9": [ | |
| 0, | |
| 1, | |
| 2, | |
| 3, | |
| 4, | |
| 5, | |
| 7, | |
| 9, | |
| 10, | |
| 11 | |
| ], | |
| "10": [ | |
| 1, | |
| 4, | |
| 5, | |
| 6, | |
| 7, | |
| 8 | |
| ], | |
| "11": [ | |
| 0, | |
| 3, | |
| 5, | |
| 7, | |
| 8, | |
| 9, | |
| 10, | |
| 11 | |
| ] | |
| }, | |
| "qk_denominator": "ori", | |
| "sephidden_pruned": false, | |
| "transform_embedding": false, | |
| "transformers_version": "4.3.2", | |
| "type_vocab_size": 2, | |
| "use_cache": true, | |
| "vocab_size": 30522 | |
| } | |