28 lines
723 B
Markdown
28 lines
723 B
Markdown
## Download Vision Encoder (Required)
|
|
|
|
This model requires an additional vision encoder `siglip2-base-p16-ve`
|
|
|
|
|
|
```bash
|
|
git lfs install
|
|
|
|
# from HuggingFace
|
|
git clone https://huggingface.co/jingyaogong/siglip2-base-p16-ve
|
|
|
|
# or from ModelScope
|
|
git clone https://modelscope.cn/models/gongjy/siglip2-base-p16-ve
|
|
```
|
|
|
|
Directory Structure
|
|
|
|
```
|
|
model/
|
|
├── siglip2-base-p16-ve/ # Vision Encoder (Required)
|
|
│ ├── config.json
|
|
│ ├── model.safetensors # ~180MB, float16
|
|
│ └── preprocessor_config.json
|
|
├── model_minimind.py # LLM structure
|
|
├── model_vlm.py # VLM structure
|
|
├── tokenizer.json
|
|
└── tokenizer_config.json
|
|
``` |