A new tutorial shows rapid Krea 2 LoRA training with recaptioning and LR tricks
shootthesound · reddit · 2026-07-29
The author shares a rapid Krea 2 LoRA training tutorial and a GitHub repo with the implementation.
The video covers captioning, adaptive learning rates, per-image and per-epoch settings, context LoRA mode, automatic mid-training recaptioning, learning-rate promotion and demotion for individual images, and real-time likeness scoring. The attached image is a demo thumbnail for “Fizgig: Rapid KREA 2 LoRA Training Demo,” emphasizing fast fine-tuning and ultra-fast SVD / KREA 2 workflow details.
The post is explicitly meant as a tutorial plus a repo link, so the value is in the concrete training workflow rather than a model announcement.
More from coding & agent
- Student builds a local coding agent on molab and beats two 7B coder baselines — S_Conradi · 2026-07-29
- Grok launches X Timeline App Builder, questioned on sub-agent support — mattshumer_ · 2026-07-29
- Buzz adds broader agent support as agensis pitches a shared workspace for humans and agents — jasonkneen · 2026-07-29
- Sanity announces an AI builders event in Brooklyn on September 9 — davidcrawshaw · 2026-07-29
- Databricks and OpenAI speakers headline San Francisco meetup on agent observability — matei_zaharia · 2026-07-29
- FastMCP name clash hides two separate projects, with Python as the canonical one — tristanbob · 2026-07-29