Diffusers
PyTorch
Safetensors
English
llama
MIO-7B-Instruct / config.json
ZenMoore's picture
Upload folder using huggingface_hub
b858d5d verified
{
"_name_or_path": "/cpfs/user/wangzekun/repos/kangz/model_hf_origin/MIO-sft/",
"architectures": [
"LlamaForCausalLM"
],
"attention_bias": false,
"bos_token_id": 1,
"eos_token_id": 2,
"hidden_act": "silu",
"hidden_size": 4096,
"image_bos_id": 64005,
"image_eos_id": 64004,
"image_newline_id": 64002,
"image_placeholder_id": 64002,
"initializer_range": 0.02,
"intermediate_size": 11008,
"is_decoder": true,
"max_position_embeddings": 4096,
"model_type": "llama",
"num_attention_heads": 32,
"num_hidden_layers": 32,
"num_key_value_heads": 4,
"num_new_special_tokens": 6,
"packing_mask": false,
"pad_token_id": 0,
"pretraining_tp": 1,
"rms_norm_eps": 1e-05,
"rope_scaling": null,
"rope_theta": 5000000.0,
"speech_bos_id": 64003,
"speech_eos_id": 64001,
"speech_placeholder_id": 64000,
"speech_vocab_size": 4096,
"text_vocab_size": 64000,
"tie_word_embeddings": false,
"torch_dtype": "float32",
"transformers_version": "4.34.0",
"unk_token_id": 0,
"use_cache": true,
"visual_vocab_size": 8192,
"vocab_size": 76294
}