Training and tweaking LoRAs for Ace-Step 1.5, the open-source music generation model
shootthesound · reddit · 2026-10-12
A Reddit user shares hands-on experience training and tweaking LoRAs for Ace-Step 1.5, the open-source music generation model, with sample results in the gallery. The post covers how to train a LoRA on your own music data to steer the model toward specific styles, and how to adjust it afterwards for better output. A practical reference for anyone customizing AI music generation.
More from Multimodal
- Training a pixel-art image gen model with synthetic data and RL via Decisions API — Angaisb_ · 2026-10-12
- 'Banana Architecture': AI-generated artwork merges giant bananas into cutaway houses — makeitrad1 · 2026-10-12
- Qwen Image editor workflow runs on 8GB VRAM, ~20s per edit — Intelligent-While146 · 2026-10-12
- Accidental 2MP render still works — SD community probes beyond the 1MP ceiling — wildmonkeywrangler · 2026-10-12
- Stunning retro-style AI image of a family picnic goes viral — Kyrannio · 2026-10-12
- Midjourney x Odyssey-3 combo called "an out-of-body experience" — Kyrannio · 2026-10-12