Nodes/FL PenguinVL/FL PenguinVL Load Model
ComfyUI Node

FL PenguinVL Load Model

A ComfyUI node in FL PenguinVL with 5 inputs and 1 output.

By filliptm·Created 5 months ago·Updated 4 months ago· 4
FL PenguinVL Load Model
    • model
    model_variant2B
    precisionauto
    deviceauto
    use_4bitfalse
    force_reloadfalse
    CategoryFL PenguinVL

    Inputs (5)

    NameTypeDefaultDescription
    model_variantCOMBO2BModel size. 2B needs ~6GB VRAM (fp16), 8B needs ~18GB VRAM (fp16)
    precisionCOMBOautoauto = fp16 on CUDA, fp32 on CPU
    deviceCOMBOautoauto selects CUDA if available
    use_4bitoptBOOLEANfalseUse 4-bit quantization to reduce VRAM (requires bitsandbytes)
    force_reloadoptBOOLEANfalseForce re-download and reload of the model

    Outputs (1)

    NameTypeDescription
    modelFL_PENGUINVL_MODEL