| { | |
| "architectures": [ | |
| "SkimformerForMaskedLM" | |
| ], | |
| "attention_head_size": 64, | |
| "attention_probs_dropout_prob": 0.1, | |
| "contextualize_2d_positions": true, | |
| "degrade_2d_positions": false, | |
| "gradient_checkpointing": false, | |
| "hidden_act": "gelu", | |
| "hidden_dropout_prob": 0.1, | |
| "hidden_layout_size": 768, | |
| "hidden_size": 768, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 3072, | |
| "layer_norm_eps": 1e-12, | |
| "max_2d_position_embeddings": 1024, | |
| "max_position_embeddings": 512, | |
| "model_type": "skimformer", | |
| "num_attention_heads": 12, | |
| "num_attention_heads_layout_encoder": 12, | |
| "num_hidden_layers": 12, | |
| "num_hidden_layers_layout_encoder": 2, | |
| "pad_token_id": 0, | |
| "skim_attention_head_size": 64, | |
| "transformers_version": "4.5.1", | |
| "type_vocab_size": 2, | |
| "use_1d_positions": false, | |
| "vocab_size": 30522 | |
| } | |