| { | |
| "activation_dropout": 0.1, | |
| "activation_function": "gelu", | |
| "architectures": [ | |
| "MBartForConditionalGeneration" | |
| ], | |
| "attention_dropout": 0.1, | |
| "bos_token_id": 64000, | |
| "d_model": 1024, | |
| "classifier_dropout": 0.0, | |
| "decoder_attention_heads": 16, | |
| "decoder_ffn_dim": 4096, | |
| "decoder_layerdrop": 0.0, | |
| "decoder_layers": 6, | |
| "dropout": 0.1, | |
| "encoder_attention_heads": 16, | |
| "encoder_ffn_dim": 4096, | |
| "encoder_layerdrop": 0.0, | |
| "encoder_layers": 6, | |
| "eos_token_id": 64001, | |
| "gradient_checkpointing": false, | |
| "init_std": 0.02, | |
| "is_encoder_decoder": true, | |
| "max_position_embeddings": 1024, | |
| "model_type": "mbart", | |
| "num_hidden_layers": 6, | |
| "pad_token_id": 0, | |
| "scale_embedding": false, | |
| "transformers_version": "4.3.2", | |
| "use_cache": true, | |
| "vocab_size": 64015, | |
| "tokenizer_class": "AlbertTokenizer" | |
| } | |