ComfyUI update breaks GGUF, INT8 and LoRAs — root cause and fixes
Occsan · reddit · 2026-10-08
A recent ComfyUI commit (0752bcb, PR #16816) crashes workflows for MiniMax, LTX and newer models with inputact TypeErrors and LoRA shape-mismatch errors. The PR removed the linearinputact() fallback and hardcoded activation args into forward(), breaking ComfyUI-GGUF, community INT8 loaders, and native LoRA loading. Fixes: manually compute SwiGLU before fc2 in comfy/ldm/minimax/model.py line 211, or roll back to commit 7d9e5a0 until upstream is patched.
More from Multimodal
- Fully AI-generated, zero human edits: Slopcore music video "KILL SWITCH" — Yamapama · 2026-10-08
- Redditors weigh --fast-disk vs RAM offloading for running large image models locally — NowThatsMalarkey · 2026-10-08
- Adaption AI pitches Invent to solve zero-data regime with AI-generated video explainer — sarahookr · 2026-10-08
- AI model Fable 5 creates 'THE LOOM', a self-portrait that's making the rounds — repligate · 2026-10-08
- Opus 5.5 redraws Ragna Crimson art via Photocraft MCP on first try — teortaxesTex · 2026-10-08
- 6 reference images and a segmented prompt: recreating live TV look with Seedance 2.5 — techhalla · 2026-10-08