Change8

v0.8.1

📦 comfyuiView on GitHub →
4 features🐛 3 fixes🔧 3 symbols

Summary

This release introduces support for Gemma 12B quantized weights and includes several optimizations for LTXAV text encoders, alongside dependency bumps and workflow fixes.

Migration Steps

  1. Bump required comfy-kitchen version.
  2. Update your pytorch to cu130 to get a performance boost.

✨ New Features

  • Support gemma 12B with quant weights.
  • Add device selection for LTXAVTextEncoderLoader.
  • Lower ltxv text encoder vram use.
  • Add memory estimation function to ltxav text encoder.

🐛 Bug Fixes

  • Fix stable release workflow not pulling latest comfy kitchen.
  • model_patcher: Remove confusing load stat.
  • ops: Fix offloading with FP8MM performance.

🔧 Affected Symbols

LTXAVTextEncoderLoadermodel_patcherops