| { | |
| "activation_dropout": 0.0, | |
| "architectures": [ | |
| "FunnelBaseModel" | |
| ], | |
| "attention_dropout": 0.1, | |
| "attention_type": "relative_shift", | |
| "block_repeats": [ | |
| 1, | |
| 1, | |
| 1 | |
| ], | |
| "block_sizes": [ | |
| 8, | |
| 8, | |
| 8 | |
| ], | |
| "d_head": 64, | |
| "d_inner": 4096, | |
| "d_model": 1024, | |
| "hidden_act": "gelu_new", | |
| "hidden_dropout": 0.1, | |
| "initializer_range": 0.1, | |
| "initializer_std": null, | |
| "layer_norm_eps": 1e-09, | |
| "max_position_embeddings": 512, | |
| "model_type": "funnel", | |
| "n_head": 16, | |
| "num_decoder_layers": 2, | |
| "pool_q_only": true, | |
| "pooling_type": "mean", | |
| "rel_attn_type": "factorized", | |
| "separate_cls": true, | |
| "truncate_seq": true, | |
| "type_vocab_size": 3, | |
| "vocab_size": 30522 | |
| } | |