| { | |
| "add_cross_attention": false, | |
| "approx_mode": "full", | |
| "architectures": [ | |
| "MraForMaskedLM" | |
| ], | |
| "attention_probs_dropout_prob": 0.0, | |
| "block_per_row": 4, | |
| "bos_token_id": 2, | |
| "dtype": "float32", | |
| "eos_token_id": 3, | |
| "hidden_act": "gelu", | |
| "hidden_dropout_prob": 0.0, | |
| "hidden_size": 16, | |
| "initial_prior_diagonal_n_blocks": 0, | |
| "initial_prior_first_n_blocks": 0, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 36, | |
| "is_decoder": false, | |
| "layer_norm_eps": 1e-05, | |
| "max_position_embeddings": 200, | |
| "model_type": "mra", | |
| "num_attention_heads": 2, | |
| "num_hidden_layers": 2, | |
| "pad_token_id": 0, | |
| "tie_word_embeddings": true, | |
| "transformers_version": "5.16.0.dev0", | |
| "type_vocab_size": 16, | |
| "vocab_size": 5 | |
| } | |