OmniLoRA: Universal LoRA trainer inside ComfyUI works on any GPU and model
pente5 · reddit · 2026-11-18
A developer releases ComfyUI-OmniLoRA, a node pack that brings a universal LoRA trainer into ComfyUI workflows, combining dataset prep, training, and monitoring in one graph.
- Broad model support: new models likely work as soon as ComfyUI supports them
- Works on any GPU with efficient CPU streaming for low-VRAM cards
- GGUF support, e.g. training on q4 quantized models
- Modular design enabling LLM dataset labeling and auto subject separation/masks
It's a beta release, available via Comfy Manager and GitHub; the author hopes the community will build a workflow library of fine-tuned parameters per model.
More from coding & agent
- Perplexity Launches Multiagent for Max Subscribers; Teams Beat Single Models on WANDR — perplexity_ai · 2026-11-18
- Proactive Agents That Schedule Their Own Work Hinge on Memory and UX, Says LangChain's Runkle — hwchase17 · 2026-11-18
- Cognition Launches Devin Bots: One Bot Leads a Team of Agents — aidangch · 2026-11-18
- 2026's 10 Best AI Agent Frameworks: Agent Adoption Doubled, 90% Now Write Data — rseroter · 2026-11-18
- Replit Livestream With BabyAGI Creator Yohei Nakajima on Building Frontier Agents — ColleenMBrady · 2026-11-18
- fx: a tiny open-source Zig coding agent with a 6.33 MiB binary that runs in the browser — evilrabbit_ · 2026-11-18