Jared Palmer runs overnight autoresearch with Devin on Modal to improve a 0.6B model

AAAzzam · x · 2026-09-19

Jared Palmer demonstrates an autoresearch workflow: he set up Devin to run training experiments overnight on Modal to produce an improved Qwen3-0.6B checkpoint. He previously open-sourced Kev-0.5B, a tiny decision model based on Qwen2.5-0.5B with a TypeSafe-compatible API that can be trained and run on a MacBook Pro, with weights and model card on GitHub.

Original post →

More from coding & agent

coding & agent channel →