model1 / masp_094_v2 /config.json
multitensor's picture
Upload folder using huggingface_hub
bbfa6f6 verified
{
"_name_or_path": "mistralai/Mistral-7B-Instruct-v0.3",
"adapter_module_name": null,
"adapter_module_path": null,
"architectures": [
"LlavaMistralForCausalLM"
],
"attention_dropout": 0.0,
"bos_token_id": 1,
"drop_path_rate": 0.0,
"eos_token_id": 2,
"freeze_mm_mlp_adapter": false,
"hidden_act": "silu",
"hidden_size": 4096,
"image_aspect_ratio": "anyres",
"image_grid_pinpoints": "[(448, 672), (672, 448)]",
"img_size": 224,
"initializer_range": 0.02,
"intermediate_size": 14336,
"lora_lr": null,
"max_num_segments": 10,
"max_position_embeddings": 32768,
"mm_hidden_size": 768,
"mm_patch_merge_type": "flat",
"mm_projector_lr": null,
"mm_projector_type": "linear",
"mm_use_patch_token": false,
"mm_use_start_end": true,
"mm_vision_select_feature": "patch",
"mm_vision_select_layer": -2,
"mm_vision_tower": "eva-vit-g",
"model_type": "llava_mistral",
"num_attention_heads": 32,
"num_hidden_layers": 32,
"num_key_value_heads": 8,
"num_query_token": 32,
"qformer_model_path": "./model/blip2_pretrained_flant5xxl.pth",
"rms_norm_eps": 1e-05,
"rope_theta": 1000000.0,
"sliding_window": null,
"tie_word_embeddings": false,
"tokenizer_model_max_length": 4096,
"tokenizer_padding_side": "right",
"torch_dtype": "bfloat16",
"transformers_version": "4.38.2",
"tune_mm_mlp_adapter": false,
"use_cache": true,
"use_mm_proj": true,
"vit_model_path": "./model/eva_vit_g.pth",
"vit_precision": "fp16",
"vocab_size": 32772
}