Training a robot character LoRA: dataset and checkpoints shared
no3us · reddit · 2026-10-27
Reddit user no3us shares a full robot character LoRA training run, releasing the dataset and saved checkpoints. The attached eval sheet compares checkpoints at different steps, noting "Later is not always better" — earlier steps can outperform longer training, a useful reference for character-LoRA work.
More from Multimodal
- Rumor: Google may unveil Fable 5.5 video model tomorrow to counter OpenAI — rand_longevity · 2026-10-27
- Fable 5.1 autonomously directs a 9-minute TSMC documentary for $19.79, zero human editing — drrickio · 2026-10-27
- EGO: a full AI-generated action short film released on Reddit — primehouse_studio · 2026-10-27
- Animating rapper EsDeeKid's Bandit in Æon Flux style with AI video — bennash · 2026-10-27
- Scale CEO turns viral sperm video into a playable game with Muse Wonder Full — alexandr_wang · 2026-10-27
- Custom ComfyUI node pushes MiniMax H3 video to 447 frames with consistent motion — No-Property3068 · 2026-10-27