twright8's picture
Trained with Unsloth
a22e879 verified
raw
history blame contribute delete
730 Bytes
{
"_name_or_path": "unsloth/Phi-3-mini-4k-instruct",
"architectures": [
"MistralForCausalLM"
],
"attention_dropout": 0.0,
"bos_token_id": 1,
"eos_token_id": 32000,
"hidden_act": "silu",
"hidden_size": 3072,
"initializer_range": 0.02,
"intermediate_size": 8192,
"max_position_embeddings": 4096,
"model_type": "mistral",
"num_attention_heads": 32,
"num_hidden_layers": 32,
"num_key_value_heads": 32,
"pad_token_id": 32000,
"rms_norm_eps": 1e-05,
"rope_scaling": null,
"rope_theta": 10000.0,
"sliding_window": 2048,
"tie_word_embeddings": false,
"torch_dtype": "bfloat16",
"transformers_version": "4.42.3",
"unsloth_version": "2024.7",
"use_cache": true,
"vocab_size": 32064
}