Mongjin/time_memory_model

1{2  "_name_or_path": "Mongjin/ft_memory_model",3  "activation_dropout": 0.0,4  "activation_function": "gelu",5  "add_bias_logits": false,6  "add_final_layer_norm": false,7  "architectures": [8    "BartForDialogueGeneration"9  ],10  "attention_dropout": 0.0,11  "author": "Heewon Jeon(madjakarta@gmail.com)",12  "bos_token_id": 1,13  "classif_dropout": 0.1,14  "classifier_dropout": 0.1,15  "d_model": 768,16  "decoder_attention_heads": 16,17  "decoder_ffn_dim": 3072,18  "decoder_layerdrop": 0.0,19  "decoder_layers": 6,20  "decoder_start_token_id": 1,21  "do_blenderbot_90_layernorm": false,22  "dropout": 0.1,23  "encoder_attention_heads": 16,24  "encoder_ffn_dim": 3072,25  "encoder_layerdrop": 0.0,26  "encoder_layers": 6,27  "eos_token_id": 1,28  "extra_pos_embeddings": 2,29  "force_bos_token_to_be_generated": false,30  "forced_eos_token_id": 1,31  "gradient_checkpointing": false,32  "id2label": {33    "0": "NEGATIVE",34    "1": "POSITIVE"35  },36  "init_std": 0.02,37  "is_encoder_decoder": true,38  "kobart_version": 2.0,39  "label2id": {40    "NEGATIVE": 0,41    "POSITIVE": 142  },43  "max_position_embeddings": 1026,44  "model_type": "bart",45  "normalize_before": false,46  "normalize_embedding": true,47  "num_hidden_layers": 6,48  "pad_token_id": 3,49  "scale_embedding": false,50  "static_position_embeddings": false,51  "tokenizer_class": "PreTrainedTokenizerFast",52  "torch_dtype": "float32",53  "transformers_version": "4.17.0",54  "use_cache": true,55  "vocab_size": 3000456}57