wanadzhar913's picture
Upload MistralForSequenceClassification
de174dc verified
raw
history blame contribute delete
847 Bytes
{
"_name_or_path": "reranker-malaysian-mistral-191M/checkpoint-27200",
"architectures": [
"MistralForSequenceClassification"
],
"attention_dropout": 0.0,
"auto_map": {
"AutoModel": "classifier.MistralForSequenceClassification"
},
"bos_token_id": 1,
"eos_token_id": 2,
"head_dim": 48,
"hidden_act": "silu",
"hidden_size": 768,
"initializer_range": 0.02,
"intermediate_size": 3072,
"max_position_embeddings": 4096,
"model_type": "mistral",
"num_attention_heads": 16,
"num_hidden_layers": 16,
"num_key_value_heads": 8,
"pad_token_id": 0,
"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": "4.43.3",
"use_cache": true,
"vocab_size": 32000
}