SteerDistill: closed-loop teacher steering keeps sub-4-bit LLM agents within 1 point of BF16
_akhaliq · x · 2026-11-13
SteerDistill quantizes LLM agents via closed-loop teacher steering: a frozen BF16 teacher guides student retries and refines guidance from execution feedback, with context distillation internalizing verified corrections so the student runs alone afterward. Results: at W4A16, average scores stay within one point of BF16 across three models and five agent benchmarks; at W2A16 it beats tested quantized baselines on Qwen3-8B. Code, paper and project page are public.
More from Research
- Speculative Self-Distillation cuts training tokens 38–69% with no accuracy loss — Azaliamirh · 2026-11-13
- tszzl: local minima are rare in high-dimensional spaces — the greatest whitepill — JensHonack · 2026-11-13
- Chronicle Bio announces first clinical trial with Prolong, using AI to match patients for ME/CFS drug — thekaransinghal · 2026-11-13
- Salesforce's A²-Bench: 500 scenarios expose 20 failure modes in multi-agent teamwork — mohitban47 · 2026-11-13
- Matrix optimizers likely within 5-10% CE of each other, researcher argues — _arohan_ · 2026-11-13
- Princeton/Stanford study: optimizer rankings reverse as batch size changes — _arohan_ · 2026-11-13