RUPunct/RUPunct_big
748k
1{2 "_name_or_path": "output_bert",3 "architectures": [4 "BertForTokenClassification"5 ],6 "attention_probs_dropout_prob": 0.1,7 "classifier_dropout": null,8 "directionality": "bidi",9 "hidden_act": "gelu",10 "hidden_dropout_prob": 0.1,11 "hidden_size": 768,12 "id2label": {13 "0": "UPPER_PERIOD",14 "1": "LOWER_PERIOD",15 "2": "UPPER_TOTAL_PERIOD",16 "3": "UPPER_COMMA",17 "4": "LOWER_COMMA",18 "5": "UPPER_TOTAL_COMMA",19 "6": "UPPER_QUESTION",20 "7": "LOWER_QUESTION",21 "8": "UPPER_TOTAL_QUESTION",22 "9": "UPPER_TIRE",23 "10": "LOWER_TIRE",24 "11": "UPPER_TOTAL_TIRE",25 "12": "UPPER_VOSKL",26 "13": "LOWER_VOSKL",27 "14": "UPPER_TOTAL_VOSKL",28 "15": "UPPER_DVOETOCHIE",29 "16": "LOWER_DVOETOCHIE",30 "17": "UPPER_TOTAL_DVOETOCHIE",31 "18": "UPPER_PERIODCOMMA",32 "19": "LOWER_PERIODCOMMA",33 "20": "UPPER_TOTAL_PERIODCOMMA",34 "21": "UPPER_DEFIS",35 "22": "LOWER_DEFIS",36 "23": "UPPER_TOTAL_DEFIS",37 "24": "UPPER_QUESTIONVOSKL",38 "25": "LOWER_QUESTIONVOSKL",39 "26": "UPPER_TOTAL_QUESTIONVOSKL",40 "27": "UPPER_MNOGOTOCHIE",41 "28": "LOWER_MNOGOTOCHIE",42 "29": "UPPER_TOTAL_MNOGOTOCHIE",43 "30": "UPPER_O",44 "31": "LOWER_O",45 "32": "UPPER_TOTAL_O"46 },47 "initializer_range": 0.02,48 "intermediate_size": 3072,49 "label2id": {50 "LOWER_COMMA": 4,51 "LOWER_DEFIS": 22,52 "LOWER_DVOETOCHIE": 16,53 "LOWER_MNOGOTOCHIE": 28,54 "LOWER_O": 31,55 "LOWER_PERIOD": 1,56 "LOWER_PERIODCOMMA": 19,57 "LOWER_QUESTION": 7,58 "LOWER_QUESTIONVOSKL": 25,59 "LOWER_TIRE": 10,60 "LOWER_VOSKL": 13,61 "UPPER_COMMA": 3,62 "UPPER_DEFIS": 21,63 "UPPER_DVOETOCHIE": 15,64 "UPPER_MNOGOTOCHIE": 27,65 "UPPER_O": 30,66 "UPPER_PERIOD": 0,67 "UPPER_PERIODCOMMA": 18,68 "UPPER_QUESTION": 6,69 "UPPER_QUESTIONVOSKL": 24,70 "UPPER_TIRE": 9,71 "UPPER_TOTAL_COMMA": 5,72 "UPPER_TOTAL_DEFIS": 23,73 "UPPER_TOTAL_DVOETOCHIE": 17,74 "UPPER_TOTAL_MNOGOTOCHIE": 29,75 "UPPER_TOTAL_O": 32,76 "UPPER_TOTAL_PERIOD": 2,77 "UPPER_TOTAL_PERIODCOMMA": 20,78 "UPPER_TOTAL_QUESTION": 8,79 "UPPER_TOTAL_QUESTIONVOSKL": 26,80 "UPPER_TOTAL_TIRE": 11,81 "UPPER_TOTAL_VOSKL": 14,82 "UPPER_VOSKL": 1283 },84 "layer_norm_eps": 1e-12,85 "max_position_embeddings": 512,86 "model_type": "bert",87 "num_attention_heads": 12,88 "num_hidden_layers": 12,89 "pad_token_id": 0,90 "pooler_fc_size": 768,91 "pooler_num_attention_heads": 12,92 "pooler_num_fc_layers": 3,93 "pooler_size_per_head": 128,94 "pooler_type": "first_token_transform",95 "position_embedding_type": "absolute",96 "torch_dtype": "float32",97 "transformers_version": "4.28.1",98 "type_vocab_size": 2,99 "use_cache": true,100 "vocab_size": 120138101}102 