YuE2 LoRA Training for New Genres Keeps Failing, Redditor Asks for Help
EuphoricTrainer311 · reddit · 2026-10-09
A redditor tried training LoRA for new music genres on YuE2 using Fill's trainer, AiToolkit and YuE2 Studio across datasets of 30-200 tracks, 50-1200 epochs/steps and varied inference settings, but only got garbled noise or results identical to no-LoRA. Acestep trains new genres fine, albeit with default audio quality issues.
More from Multimodal
- RTX VSR Studio: open-source Windows app upscales 2-hour films with frame-exact resume — garionhk · 2026-10-09
- Claude Motion generates code-based editable animations, no video model — claudeai · 2026-10-09
- Open-source local image library asks: how do you find an old generation by its settings? — shivam_dewan · 2026-10-09
- Iris-3B: a 3B pixel-space text-to-image model with no VAE, fully open under Apache 2.0 — zhenjun_zhao · 2026-10-09
- Voyager launches: an open harness that plugs frontier models into creative tools — testingcatalog · 2026-10-09
- Voyager launches as an open harness driving Opus, Astra and DeepSeek across AE, Blender and more — HeyAmit_ · 2026-10-09