|
{ |
|
"_name_or_path": "cross-encoder/nli-distilroberta-base", |
|
"_num_labels": 13, |
|
"architectures": [ |
|
"RobertaForSequenceClassification" |
|
], |
|
"attention_probs_dropout_prob": 0.1, |
|
"bos_token_id": 0, |
|
"classifier_dropout": null, |
|
"eos_token_id": 2, |
|
"gradient_checkpointing": false, |
|
"hidden_act": "gelu", |
|
"hidden_dropout_prob": 0.1, |
|
"hidden_size": 768, |
|
"id2label": { |
|
"0": "A lot of people believe [MSK1]. Therefore, [MSK1] must be true.", |
|
"1": "Either [MSK1] or [MSK2] is true.", |
|
"2": "If [MSK1] is true, then [MSK2] is true. [MSK2] is true. Therefore, [MSK1] is true.", |
|
"3": "It is claimed that [MSK1] implies [MSK2], whereas [MSK1] is unrelated to [MSK2].", |
|
"4": "[MSK1] claims that [MSK2]. [MSK1] are experts in the field concerning [MSK2]. Therefore, [MSK2] should be believed.", |
|
"5": "[MSK1] has attribute [MSK2]. [MSK1] is a subset of [MSK3]. Therefore, all [MSK3] has attribute [MSK2].", |
|
"6": "[MSK1] is claiming [MSK2]. [MSK1] is a moron. Therefore, [MSK2] is not true.", |
|
"7": "[MSK1] is made without evidence. In place of evidence, emotion is used to convince the interlocutor that [MSK1] is true.", |
|
"8": "[MSK1] is true because of [MSK2]. [MSK2] is true because of [MSK1].", |
|
"9": "[MSK1] is used to mean [MSK2] in the premise. [MSK1] is used to mean [MSK3] in the conclusion.", |
|
"10": "[MSK1] knows [MSK2] is incorrect. [MSK1] still claim that [MSK2] is correct using an incorrect argument.", |
|
"11": "[MSK1] makes claim [MSK2]. [MSK3] restates [MSK2] (in a distorted way). [MSK3] attacks the distorted version of [MSK2]. Therefore, [MSK2] is false.", |
|
"12": "[MSK1] occurred, then [MSK2] occurred. Therefore, [MSK1] caused [MSK2]." |
|
}, |
|
"initializer_range": 0.02, |
|
"intermediate_size": 3072, |
|
"label2id": { |
|
"A lot of people believe [MSK1]. Therefore, [MSK1] must be true.": 0, |
|
"Either [MSK1] or [MSK2] is true.": 1, |
|
"If [MSK1] is true, then [MSK2] is true. [MSK2] is true. Therefore, [MSK1] is true.": 2, |
|
"It is claimed that [MSK1] implies [MSK2], whereas [MSK1] is unrelated to [MSK2].": 3, |
|
"[MSK1] claims that [MSK2]. [MSK1] are experts in the field concerning [MSK2]. Therefore, [MSK2] should be believed.": 4, |
|
"[MSK1] has attribute [MSK2]. [MSK1] is a subset of [MSK3]. Therefore, all [MSK3] has attribute [MSK2].": 5, |
|
"[MSK1] is claiming [MSK2]. [MSK1] is a moron. Therefore, [MSK2] is not true.": 6, |
|
"[MSK1] is made without evidence. In place of evidence, emotion is used to convince the interlocutor that [MSK1] is true.": 7, |
|
"[MSK1] is true because of [MSK2]. [MSK2] is true because of [MSK1].": 8, |
|
"[MSK1] is used to mean [MSK2] in the premise. [MSK1] is used to mean [MSK3] in the conclusion.": 9, |
|
"[MSK1] knows [MSK2] is incorrect. [MSK1] still claim that [MSK2] is correct using an incorrect argument.": 10, |
|
"[MSK1] makes claim [MSK2]. [MSK3] restates [MSK2] (in a distorted way). [MSK3] attacks the distorted version of [MSK2]. Therefore, [MSK2] is false.": 11, |
|
"[MSK1] occurred, then [MSK2] occurred. Therefore, [MSK1] caused [MSK2].": 12 |
|
}, |
|
"layer_norm_eps": 1e-05, |
|
"max_position_embeddings": 514, |
|
"model_type": "roberta", |
|
"num_attention_heads": 12, |
|
"num_hidden_layers": 6, |
|
"pad_token_id": 1, |
|
"position_embedding_type": "absolute", |
|
"problem_type": "single_label_classification", |
|
"torch_dtype": "float32", |
|
"transformers_version": "4.41.2", |
|
"type_vocab_size": 1, |
|
"use_cache": true, |
|
"vocab_size": 50265 |
|
} |
|
|