| { | |
| "activation_dropout": 0.0, | |
| "activation_function": "gelu", | |
| "add_bias_logits": false, | |
| "add_final_layer_norm": false, | |
| "architectures": [ | |
| "LongformerBartWithDoctypeForConditionalGeneration" | |
| ], | |
| "attention_dropout": 0.1, | |
| "attention_probs_dropout_prob": 0.1, | |
| "attention_window": [ | |
| 64, | |
| 64, | |
| 64, | |
| 64 | |
| ], | |
| "author": "Heewon Jeon(madjakarta@gmail.com)", | |
| "bos_token_id": 0, | |
| "classif_dropout": 0.1, | |
| "classifier_dropout": 0.1, | |
| "d_model": 512, | |
| "decoder_attention_heads": 4, | |
| "decoder_ffn_dim": 2048, | |
| "decoder_layerdrop": 0.0, | |
| "decoder_layers": 4, | |
| "decoder_start_token_id": 1, | |
| "do_blenderbot_90_layernorm": false, | |
| "doc_type_size": 4, | |
| "dropout": 0.1, | |
| "encoder_attention_heads": 4, | |
| "encoder_ffn_dim": 2048, | |
| "encoder_layerdrop": 0.0, | |
| "encoder_layers": 4, | |
| "eos_token_id": 1, | |
| "extra_pos_embeddings": 2, | |
| "force_bos_token_to_be_generated": false, | |
| "forced_eos_token_id": 1, | |
| "gradient_checkpointing": false, | |
| "init_std": 0.02, | |
| "is_encoder_decoder": true, | |
| "kobart_version": 1.0, | |
| "max_position_embeddings": 2048, | |
| "max_target_positions": 2048, | |
| "model_type": "bart", | |
| "normalize_before": false, | |
| "normalize_embedding": true, | |
| "num_hidden_layers": 6, | |
| "pad_token_id": 3, | |
| "scale_embedding": false, | |
| "static_position_embeddings": false, | |
| "tokenizer_class": "PreTrainedTokenizerFast", | |
| "torch_dtype": "float32", | |
| "transformers_version": "4.11.0", | |
| "use_cache": true, | |
| "vocab_size": 30000 | |
| } | |