| { | |
| "attention_probs_dropout_prob": 0.1, | |
| "cell": {}, | |
| "emb_size": 256, | |
| "fix_config": { | |
| "sample_hidden_size": 256, | |
| "sample_intermediate_sizes": [ | |
| 480, | |
| 480, | |
| 480, | |
| 480 | |
| ], | |
| "sample_layer_num": 4, | |
| "sample_num_attention_heads": [ | |
| 12, | |
| 12, | |
| 12, | |
| 12 | |
| ], | |
| "sample_qkv_sizes": [ | |
| 192, | |
| 192, | |
| 192, | |
| 192 | |
| ] | |
| }, | |
| "hidden_act": "gelu", | |
| "hidden_dropout_prob": 0.1, | |
| "hidden_size": 256, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 480, | |
| "layer_norm_eps": 1e-12, | |
| "max_position_embeddings": 512, | |
| "num_attention_heads": 12, | |
| "num_hidden_layers": 4, | |
| "pre_trained": "", | |
| "qkv_size": 192, | |
| "structure": [], | |
| "training": "", | |
| "type_vocab_size": 2, | |
| "vocab_size": 30522 | |
| } | |