SENPAI Auto-Research Agent Writes Kernels, Cuts Training Time 22% (26h to 22h)
morgymcg · x · 2026-09-22
The author's team used W&B's open-source autonomous research agent SENPAI to run an automated ML research loop: the agent researched and wrote two optimizations on its own — a SliceAttention kernel (for Transolver) and a BatchedMuon implementation — cutting a full training run by 22% (26h → 22h, 4 hours saved) with no significant change in error or quality.
SENPAI is a semi-autonomous research harness: an Advisor proposes and reviews experiments, and Student agents each implement one assigned experiment, iterating on observability data. The repo ships a full LLM-training-optimization guide, paper, and tooling to reproduce the workflow.
More from coding & agent
- Moonshot Launches Kimi Browser Extension: Record Steps Once, Agent Replays Forever — Kimi_Moonshot · 2026-09-22
- Jimothy distills LLM queries into tiny task-specific classifiers running 10-20x faster — flngr · 2026-09-22
- What Cloudflare can't do: D1 caps at 10GB, is single-threaded, and no real Postgres — Paimaamu · 2026-09-22
- User Vibecodes Tampermonkey Scripts to Patch Long-Ignored UX Bugs in University Systems — craig_macdonald · 2026-09-22
- Speaking UI layouts into existence with Jev, shadcn and a local transcription model — _AustinCalvert_ · 2026-09-22
- ai-copywriter hits 1k stars: an agent skill that writes converting copy with zero AI tells — tom_doerr · 2026-09-22