newly_fine_tuned_bert / config.json
pingkeest's picture
pingkeest/newly_fine_tuned_bert
ec7a50c verified
raw
history blame
3.55 kB
{
"_name_or_path": "bert-base-uncased",
"architectures": [
"BertForSequenceClassification"
],
"attention_probs_dropout_prob": 0.1,
"classifier_dropout": null,
"gradient_checkpointing": false,
"hidden_act": "gelu",
"hidden_dropout_prob": 0.1,
"hidden_size": 768,
"id2label": {
"0": "1000",
"1": "1001",
"2": "1002",
"3": "1004",
"4": "1005",
"5": "1006",
"6": "1007",
"7": "1008",
"8": "1009",
"9": "1010",
"10": "1011",
"11": "1012",
"12": "1013",
"13": "1014",
"14": "1015",
"15": "1016",
"16": "1017",
"17": "1018",
"18": "1019",
"19": "1020",
"20": "1021",
"21": "1022",
"22": "1023",
"23": "1024",
"24": "1025",
"25": "1026",
"26": "1027",
"27": "1028",
"28": "1029",
"29": "1030",
"30": "1031",
"31": "1032",
"32": "1033",
"33": "1034",
"34": "1035",
"35": "1036",
"36": "1037",
"37": "1038",
"38": "1039",
"39": "1040",
"40": "1041",
"41": "1042",
"42": "1043",
"43": "1044",
"44": "1045",
"45": "1046",
"46": "1047",
"47": "1048",
"48": "1049",
"49": "1050",
"50": "1051",
"51": "1052",
"52": "1053",
"53": "1054",
"54": "1055",
"55": "1056",
"56": "1057",
"57": "1058",
"58": "1059",
"59": "1060",
"60": "1061",
"61": "1062",
"62": "1063",
"63": "1064",
"64": "1065",
"65": "1066",
"66": "1067",
"67": "1068",
"68": "1069",
"69": "1070",
"70": "1071",
"71": "1072",
"72": "1073",
"73": "1074",
"74": "1075",
"75": "1076",
"76": "1077"
},
"initializer_range": 0.02,
"intermediate_size": 3072,
"label2id": {
"1000": 0,
"1001": 1,
"1002": 2,
"1004": 3,
"1005": 4,
"1006": 5,
"1007": 6,
"1008": 7,
"1009": 8,
"1010": 9,
"1011": 10,
"1012": 11,
"1013": 12,
"1014": 13,
"1015": 14,
"1016": 15,
"1017": 16,
"1018": 17,
"1019": 18,
"1020": 19,
"1021": 20,
"1022": 21,
"1023": 22,
"1024": 23,
"1025": 24,
"1026": 25,
"1027": 26,
"1028": 27,
"1029": 28,
"1030": 29,
"1031": 30,
"1032": 31,
"1033": 32,
"1034": 33,
"1035": 34,
"1036": 35,
"1037": 36,
"1038": 37,
"1039": 38,
"1040": 39,
"1041": 40,
"1042": 41,
"1043": 42,
"1044": 43,
"1045": 44,
"1046": 45,
"1047": 46,
"1048": 47,
"1049": 48,
"1050": 49,
"1051": 50,
"1052": 51,
"1053": 52,
"1054": 53,
"1055": 54,
"1056": 55,
"1057": 56,
"1058": 57,
"1059": 58,
"1060": 59,
"1061": 60,
"1062": 61,
"1063": 62,
"1064": 63,
"1065": 64,
"1066": 65,
"1067": 66,
"1068": 67,
"1069": 68,
"1070": 69,
"1071": 70,
"1072": 71,
"1073": 72,
"1074": 73,
"1075": 74,
"1076": 75,
"1077": 76
},
"layer_norm_eps": 1e-12,
"max_position_embeddings": 512,
"model_type": "bert",
"num_attention_heads": 12,
"num_hidden_layers": 12,
"pad_token_id": 0,
"position_embedding_type": "absolute",
"problem_type": "multi_label_classification",
"torch_dtype": "float32",
"transformers_version": "4.45.2",
"type_vocab_size": 2,
"use_cache": true,
"vocab_size": 30522
}