lym00/HunyuanVideo-Avatar-GGUF

tencent/HunyuanVideo-Avatar GGUF Conversion

This repository contains a direct GGUF conversion of the tencent/HunyuanVideo-Avatar model, originally sourced from tencent/HunyuanVideo-Avatar.

All quantized versions were created from the base model ckpts/hunyuan-video-t2v-720p/transformers/mp_rank_00_model_states.pt using the conversion scripts provided by city96, available at the ComfyUI-GGUF GitHub repository.

The process involved first converting the pickletensors model to a BF16 GGUF, then quantizing it, and finally applying the 5D fix.

Usage

As this is a quantized model not a finetune, all the same restrictions/original license terms still apply.

• The model files are compatible with the ComfyUI-GGUF custom node.
• Place the model files in the directory:
  ComfyUI/models/unet
• For detailed installation instructions, please refer to the ComfyUI-GGUF GitHub repository.

Additional Resources

• The VAE can be downloaded from Kijai’s repository on Hugging Face.

Reference

• For an overview of quantization types, please see the LLaMA 3 8B Scoreboard quantization chart.