Metadata-Version: 2.1
Name: comfyui-turboquant
Version: 0.0.1
Summary: TQ3 KV cache compression for ComfyUI reducing attention KV cache VRAM by ~4.5x using 3-bit Lloyd-Max quantization with Fast Walsh-Hadamard Transform decorrelation.
Home-page: https://github.com/Scottcjn/ComfyUI-TurboQuant
Requires-Python: >=3.10
