← All repos

yuvhaim-gif/LLM_InSight

View on GitHub ↗ ★ 5 Python

This is my personal home rig for serious LLM experimentation. I built it to test models head-to-head, create custom evaluation rubrics, automatically improve prompts based on the previous run’s results, and generate high-quality synthetic training data. Everything runs locally first (Ollama by default), with optional cloud support. logged locally.

4 mentions on Hacker News by 1 person: 4 stories, 0 comments · first 2026-05-04 · latest 2026-06-15