LLM-as-a-Coach turns judge feedback into transferable experiential knowledge
iScienceLuvr · x · 2026-07-21
This is a duplicate share of the paper “LLM-as-a-Coach: Experiential Learning for Non-Verifiable Tasks.”
It proposes Experiential Learning (EL), which converts LLM-as-a-Judge style feedback into an LLM-as-a-Coach. Rather than reducing evaluation to a scalar reward, the coach distills its judgment into transferable experiential knowledge that conditions a teacher model and is absorbed by the policy via on-policy context distillation.
The paper’s main claim is that this richer feedback signal improves open-ended, non-verifiable tasks, generalizes better out of distribution, and reduces reward hacking compared with rubric-based RL.
Related event: Microsoft Proposes LLM-as-a-Coach for Non-Verifiable Tasks(4 posts)→
More from Research
- Open-source decision models sold their own base in Red Alert 2 — then learned until no human could beat them — heathercmiller · 2026-10-27
- DeepMind paper: LLMs can't sample randomly on their own — pass random numbers from code instead — rohanpaul_ai · 2026-10-27
- World Model Assigns Weight and Friction to Objects, Then Simulates with a Physics Engine — jonstephens85 · 2026-10-27
- Token Collapse: Study Finds Graph Soft Prompting Tokens in LLMs Fail to Encode Structure — Sapienza · 2026-10-27
- Mathathon 2.0 launches with Terence Tao's SAIR, $1.5M in compute for teams — _sathvikr · 2026-10-27
- The 'sweet lesson' is real: two kinds of ML problems need opposite strategies — mike64_t · 2026-10-27