Please re-download int8 version. I have uploaded much faster, proper version.<hr /><h3 id="versions">Versions</h3><ul><li>int8: recommended. Fast, accurate, compatible with almost any GPU.</li><li>mxfp8: added for comparison. In theory (and according to nVidia PR) should be more accurate than int8, but in practice I was not able to spot any definitive advantages. A bit slower than int8, but still faster than original bf16. Compatible only with RTX 50xx series (Blackwell).</li></ul><h3 id="performance-on-my-setup">Performance on my setup</h3><ul><li>original bf16 (baseline): 2.20 it/s +0%</li><li>int8: 3.23 it/s +46%</li><li>int8 + torch compile (comfy core): 3.59 it/s +63%</li><li>mxfp8: 2.58 it/s +17%</li></ul><hr />This is high quality int8 quantized version of WAI-ANIMA model. It retains ~90% of original model quality, but uses about 50% less VRAM and also runs faster on almost any nVidia GPU (AMD not tested). Nice trade-off, especially for low-end GPUs.Can be used as a drop-in replacement for original model in latest ComfyUI, no custom nodes required. If you have troubles running the model make sure that you updated both ComfyUI itself and its dependencies (e.g.<code>pip install -U -r requirements.txt</code> on manual linux install).Converted to int8 / mxfp8 using <a target="_blank" rel="ugc" href="https://github.com/silveroxides/convert_to_quant">convert_to_quant</a> script.

WAI-ANIMA int8 / mxfp8

int8, ConvRot with dynamic group size, rowwise, SVDPotentially better quality than ConvRot with fixed group size, but reported to be slower by some users.

Files with SHA256: d5c7c989134e32a5e9d9b6f4e500af1e1efed2da1f2c2ffaf3dc5593778c4ac5