{ "activation_dropout": 0.0, "activation_function": "gelu", "architectures": [ "MBartForConditionalGeneration" ], "attention_dropout": 0.0, "bos_token_id": 0, "classifier_dropout": 0.0, "d_model": 640, "decoder_attention_heads": 16, "decoder_ffn_dim": 2560, "decoder_layerdrop": 0.0, "decoder_layers": 12, "dropout": 0.1, "early_stopping": false, "encoder_attention_heads": 16, "encoder_ffn_dim": 2560, "encoder_layerdrop": 0.0, "encoder_layers": 12, "eos_token_id": 2, "forced_eos_token_id": 2, "init_std": 0.02, "is_encoder_decoder": true, "max_position_embeddings": 1024, "model_type": "mbart", "num_beams": 5, "max_new_tokens": 127, "num_hidden_layers": 12, "output_hidden_states": false, "output_past": false, "pad_token_id": 1, "scale_embedding": false, "torch_dtype": "bfloat16", "transformers_version": "4.39.0", "use_cache": true, "vocab_size": 250055 }