| { | |
| "architectures": [ | |
| "LlamaForSequenceClassification" | |
| ], | |
| "attention_bias": false, | |
| "attention_dropout": 0.0, | |
| "bos_token_id": 1, | |
| "eos_token_id": 2, | |
| "head_dim": 64, | |
| "hidden_act": "silu", | |
| "hidden_size": 2048, | |
| "id2label": { | |
| "0": "True_Correct:NA", | |
| "1": "False_Neither:NA", | |
| "2": "True_Neither:NA", | |
| "3": "False_Misconception:Incomplete", | |
| "4": "False_Misconception:Additive", | |
| "5": "False_Misconception:Duplication", | |
| "6": "False_Misconception:Subtraction", | |
| "7": "False_Misconception:Wrong_term", | |
| "8": "False_Misconception:Positive", | |
| "9": "False_Misconception:Wrong_fraction", | |
| "10": "False_Misconception:Inversion", | |
| "11": "False_Misconception:Irrelevant", | |
| "12": "False_Misconception:Denominator-only_change", | |
| "13": "False_Misconception:Mult", | |
| "14": "False_Misconception:Adding_across", | |
| "15": "False_Misconception:Whole_numbers_larger", | |
| "16": "False_Misconception:Unknowable", | |
| "17": "False_Misconception:Wrong_Fraction", | |
| "18": "False_Misconception:WNB", | |
| "19": "False_Correct:NA", | |
| "20": "False_Misconception:SwapDividend", | |
| "21": "False_Misconception:Scale", | |
| "22": "True_Misconception:Tacking", | |
| "23": "False_Misconception:Not_variable", | |
| "24": "False_Misconception:Tacking", | |
| "25": "False_Misconception:Firstterm", | |
| "26": "False_Misconception:Multiplying_by_4", | |
| "27": "False_Misconception:Adding_terms", | |
| "28": "True_Misconception:Irrelevant", | |
| "29": "False_Misconception:FlipChange", | |
| "30": "False_Misconception:Division", | |
| "31": "False_Misconception:Interior", | |
| "32": "False_Misconception:Definition", | |
| "33": "True_Misconception:Additive", | |
| "34": "False_Misconception:Longer_is_bigger", | |
| "35": "False_Misconception:Base_rate", | |
| "36": "False_Misconception:Ignores_zeroes", | |
| "37": "False_Misconception:Inverse_operation", | |
| "38": "False_Misconception:Certainty", | |
| "39": "True_Misconception:Shorter_is_bigger", | |
| "40": "True_Misconception:Firstterm", | |
| "41": "True_Misconception:Wrong_term", | |
| "42": "True_Misconception:Mult", | |
| "43": "True_Misconception:SwapDividend", | |
| "44": "True_Misconception:WNB", | |
| "45": "True_Misconception:Incomplete", | |
| "46": "False_Misconception:Wrong_Operation", | |
| "47": "False_Misconception:Shorter_is_bigger", | |
| "48": "True_Misconception:Wrong_fraction", | |
| "49": "True_Misconception:Duplication", | |
| "50": "True_Misconception:Division", | |
| "51": "False_Misconception:Incorrect_equivalent_fraction_addition", | |
| "52": "True_Misconception:Inversion", | |
| "53": "True_Misconception:FlipChange", | |
| "54": "True_Misconception:Denominator-only_change", | |
| "55": "True_Misconception:Multiplying_by_4", | |
| "56": "True_Misconception:Definition", | |
| "57": "True_Misconception:Subtraction", | |
| "58": "True_Misconception:Positive", | |
| "59": "True_Misconception:Incorrect_equivalent_fraction_addition" | |
| }, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 5632, | |
| "label2id": { | |
| "False_Correct:NA": 19, | |
| "False_Misconception:Adding_across": 14, | |
| "False_Misconception:Adding_terms": 27, | |
| "False_Misconception:Additive": 4, | |
| "False_Misconception:Base_rate": 35, | |
| "False_Misconception:Certainty": 38, | |
| "False_Misconception:Definition": 32, | |
| "False_Misconception:Denominator-only_change": 12, | |
| "False_Misconception:Division": 30, | |
| "False_Misconception:Duplication": 5, | |
| "False_Misconception:Firstterm": 25, | |
| "False_Misconception:FlipChange": 29, | |
| "False_Misconception:Ignores_zeroes": 36, | |
| "False_Misconception:Incomplete": 3, | |
| "False_Misconception:Incorrect_equivalent_fraction_addition": 51, | |
| "False_Misconception:Interior": 31, | |
| "False_Misconception:Inverse_operation": 37, | |
| "False_Misconception:Inversion": 10, | |
| "False_Misconception:Irrelevant": 11, | |
| "False_Misconception:Longer_is_bigger": 34, | |
| "False_Misconception:Mult": 13, | |
| "False_Misconception:Multiplying_by_4": 26, | |
| "False_Misconception:Not_variable": 23, | |
| "False_Misconception:Positive": 8, | |
| "False_Misconception:Scale": 21, | |
| "False_Misconception:Shorter_is_bigger": 47, | |
| "False_Misconception:Subtraction": 6, | |
| "False_Misconception:SwapDividend": 20, | |
| "False_Misconception:Tacking": 24, | |
| "False_Misconception:Unknowable": 16, | |
| "False_Misconception:WNB": 18, | |
| "False_Misconception:Whole_numbers_larger": 15, | |
| "False_Misconception:Wrong_Fraction": 17, | |
| "False_Misconception:Wrong_Operation": 46, | |
| "False_Misconception:Wrong_fraction": 9, | |
| "False_Misconception:Wrong_term": 7, | |
| "False_Neither:NA": 1, | |
| "True_Correct:NA": 0, | |
| "True_Misconception:Additive": 33, | |
| "True_Misconception:Definition": 56, | |
| "True_Misconception:Denominator-only_change": 54, | |
| "True_Misconception:Division": 50, | |
| "True_Misconception:Duplication": 49, | |
| "True_Misconception:Firstterm": 40, | |
| "True_Misconception:FlipChange": 53, | |
| "True_Misconception:Incomplete": 45, | |
| "True_Misconception:Incorrect_equivalent_fraction_addition": 59, | |
| "True_Misconception:Inversion": 52, | |
| "True_Misconception:Irrelevant": 28, | |
| "True_Misconception:Mult": 42, | |
| "True_Misconception:Multiplying_by_4": 55, | |
| "True_Misconception:Positive": 58, | |
| "True_Misconception:Shorter_is_bigger": 39, | |
| "True_Misconception:Subtraction": 57, | |
| "True_Misconception:SwapDividend": 43, | |
| "True_Misconception:Tacking": 22, | |
| "True_Misconception:WNB": 44, | |
| "True_Misconception:Wrong_fraction": 48, | |
| "True_Misconception:Wrong_term": 41, | |
| "True_Neither:NA": 2 | |
| }, | |
| "max_position_embeddings": 2048, | |
| "mlp_bias": false, | |
| "model_type": "llama", | |
| "num_attention_heads": 32, | |
| "num_hidden_layers": 22, | |
| "num_key_value_heads": 4, | |
| "pad_token_id": 2, | |
| "pretraining_tp": 1, | |
| "problem_type": "single_label_classification", | |
| "rms_norm_eps": 1e-05, | |
| "rope_scaling": null, | |
| "rope_theta": 10000.0, | |
| "tie_word_embeddings": false, | |
| "torch_dtype": "bfloat16", | |
| "transformers_version": "4.54.0", | |
| "use_cache": true, | |
| "vocab_size": 32000 | |
| } | |