ComfyUI Node
🖼️ Vision Model Loader (Transformers)
A ComfyUI node in 🤖 GGUF-VLM/🖼️ Vision Models with 6 inputs and 1 output.
🖼️ Vision Model Loader (Transformers)
- model_config
◄modelHuihui-Qwen3-VL-4B-Instruct-abliterated►
◄quantizationnone►
◄attentionflash_attention_2►
◄keep_model_loadedfalse►
◄min_pixels200704►
◄max_pixels1003520►
Category🤖 GGUF-VLM/🖼️ Vision Models
Inputs (6)
| Name | Type | Default | Description |
|---|---|---|---|
| model | COMBO | Huihui-Qwen3-VL-4B-Instruct-abliterated | 选择 Qwen3-VL Abliterated 模型 |
| quantization | COMBO | none | 量化类型 |
| attention | COMBO | flash_attention_2 | 注意力机制实现（Qwen3-VL 推荐: flash_attention_2） |
| keep_model_loaded | BOOLEAN | false | 推理后是否保持模型加载在内存中 |
| min_pixels | INT | 2007043136–12845056 | 最小像素数 |
| max_pixels | INT | 10035203136–12845056 | 最大像素数 |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| model_config | TRANSFORMERS_MODEL | — |