QuantumLearningMachines/qlm-map-classifier
09
1{2 "_name_or_path": "microsoft/deberta-v3-base",3 "architectures": [4 "DebertaV2ForSequenceClassification"5 ],6 "attention_probs_dropout_prob": 0.1,7 "hidden_act": "gelu",8 "hidden_dropout_prob": 0.1,9 "hidden_size": 768,10 "id2label": {11 "0": "Incomplete",12 "1": "Additive",13 "2": "Duplication",14 "3": "Wrong_Fraction",15 "4": "Subtraction",16 "5": "Positive",17 "6": "Wrong_term",18 "7": "Irrelevant",19 "8": "Inversion",20 "9": "Mult",21 "10": "Denominator-only_change",22 "11": "Whole_numbers_larger",23 "12": "Adding_across",24 "13": "WNB",25 "14": "Tacking",26 "15": "Unknowable",27 "16": "SwapDividend",28 "17": "Scale",29 "18": "Not_variable",30 "19": "Firstterm",31 "20": "Adding_terms",32 "21": "Multiplying_by_4",33 "22": "FlipChange",34 "23": "Division",35 "24": "Definition",36 "25": "Interior",37 "26": "Longer_is_bigger",38 "27": "Ignores_zeroes",39 "28": "Shorter_is_bigger",40 "29": "Base_rate",41 "30": "Inverse_operation",42 "31": "Certainty",43 "32": "Incorrect_equivalent_fraction_addition",44 "33": "Wrong_Operation"45 },46 "initializer_range": 0.02,47 "intermediate_size": 3072,48 "label2id": {49 "Adding_across": 12,50 "Adding_terms": 20,51 "Additive": 1,52 "Base_rate": 29,53 "Certainty": 31,54 "Definition": 24,55 "Denominator-only_change": 10,56 "Division": 23,57 "Duplication": 2,58 "Firstterm": 19,59 "FlipChange": 22,60 "Ignores_zeroes": 27,61 "Incomplete": 0,62 "Incorrect_equivalent_fraction_addition": 32,63 "Interior": 25,64 "Inverse_operation": 30,65 "Inversion": 8,66 "Irrelevant": 7,67 "Longer_is_bigger": 26,68 "Mult": 9,69 "Multiplying_by_4": 21,70 "Not_variable": 18,71 "Positive": 5,72 "Scale": 17,73 "Shorter_is_bigger": 28,74 "Subtraction": 4,75 "SwapDividend": 16,76 "Tacking": 14,77 "Unknowable": 15,78 "WNB": 13,79 "Whole_numbers_larger": 11,80 "Wrong_Fraction": 3,81 "Wrong_Operation": 33,82 "Wrong_term": 683 },84 "layer_norm_eps": 1e-07,85 "legacy": true,86 "max_position_embeddings": 512,87 "max_relative_positions": -1,88 "model_type": "deberta-v2",89 "norm_rel_ebd": "layer_norm",90 "num_attention_heads": 12,91 "num_hidden_layers": 12,92 "pad_token_id": 0,93 "pooler_dropout": 0,94 "pooler_hidden_act": "gelu",95 "pooler_hidden_size": 768,96 "pos_att_type": [97 "p2c",98 "c2p"99 ],100 "position_biased_input": false,101 "position_buckets": 256,102 "relative_attention": true,103 "share_att_key": true,104 "torch_dtype": "float32",105 "transformers_version": "4.48.0",106 "type_vocab_size": 0,107 "vocab_size": 128100108}109 