DFlow2 2B T2I model distilled from 50 to 4 steps with eDMD + LADD LoRA
TimeTruth2490 · reddit · 2026-11-22
A community developer distilled the indie 2B text-to-image model DFlow2 (default 50 steps, CFG 2.5) down to 4 steps at CFG 1 using eDMD + LADD, releasing a distillation LoRA plus custom ComfyUI nodes with both Torch and MLX workflows on Hugging Face. The author says DFlow2 is impressive for its size and worth watching as it develops.
More from Multimodal
- ComfyUI launches model usage rankings: Qwen Image 2.1 tops generations at 16% — cremebait15 · 2026-11-22
- pi-H3, a flow-matching video model built on MiniMax-H3, trends on Hugging Face — Lakonik · 2026-11-22
- GPT 6.2 Astra composes Beethoven-style scherzo, with all 604 key strikes from the score — imjustnewatai · 2026-11-22
- Musk demos real-time AI video generation on half of one AI4 TRIP accelerator — jamesdouma · 2026-11-22
- RAEA engine lets AI agents create videos, 3D scenes and games with React — Promptmethus · 2026-11-22
- Do adaptive optimizers still work? Reddit users doubt it for flow-matching models — Inevitable_Pen9043 · 2026-11-22