tiny-mistral / config.json
xshubhamx's picture
Training in progress, epoch 0
256519c verified
raw
history blame
1.56 kB
{
"_name_or_path": "openaccess-ai-collective/tiny-mistral",
"architectures": [
"MistralForSequenceClassification"
],
"attention_dropout": 0.0,
"bos_token_id": 1,
"dropout_p": 0.1,
"eos_token_id": 2,
"hidden_act": "silu",
"hidden_size": 512,
"id2label": {
"0": "Issue",
"1": "Court Discourse",
"2": "Conclusion",
"3": "Precedent Analysis",
"4": "Section Analysis",
"5": "Argument by Petitioner",
"6": "Fact",
"7": "Argument by Respondent",
"8": "Ratio",
"9": "Appellant",
"10": "Respondent",
"11": "Argument by Appellant",
"12": "Petitioner",
"13": "Judge",
"14": "Argument by Defendant"
},
"initializer_range": 0.02,
"intermediate_size": 14336,
"label2id": {
"Appellant": 9,
"Argument by Appellant": 11,
"Argument by Defendant": 14,
"Argument by Petitioner": 5,
"Argument by Respondent": 7,
"Conclusion": 2,
"Court Discourse": 1,
"Fact": 6,
"Issue": 0,
"Judge": 13,
"Petitioner": 12,
"Precedent Analysis": 3,
"Ratio": 8,
"Respondent": 10,
"Section Analysis": 4
},
"max_position_embeddings": 32768,
"model_type": "mistral",
"num_attention_heads": 16,
"num_hidden_layers": 8,
"num_key_value_heads": 4,
"pad_token_id": 32000,
"problem_type": "single_label_classification",
"rms_norm_eps": 1e-05,
"rope_theta": 10000.0,
"sliding_window": 4096,
"tie_word_embeddings": false,
"torch_dtype": "float32",
"transformers_version": "4.39.3",
"use_cache": true,
"vocab_size": 32001
}