AI Paper Atlas

Research map / Reinforcement learning / RL from feedback

Auto-regressive In-context Demonstration Selection

Yunzhe Qi, Sirui Chen, Jiaru Zou, Jingrui He

ICML 2026 · Reinforcement learning › RL from feedback · approach: RL for LLM reasoning

Read the paper

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add EOS perturbation lambda=0.01 (ablation)AutoSelect · Winowhyaccuracy: 0.641 -> 0.657Improved
add EOS perturbation lambda=0.01 (ablation)AutoSelect · AQuAaccuracy: 0.379 -> 0.395Improved
add Gaussian noise std 0.1 to rewards (ablation)AutoSelect · Winowhyaccuracy: 0.657 -> 0.628Hurt
add Gaussian noise std 0.1 to rewards (ablation)AutoSelect · Trecaccuracy: 0.393 -> 0.368Hurt
increase aggregation size from 3 to 10 (ablation)AutoSelect · Trecaccuracy: 0.379 -> 0.428Improved
increase aggregation size from 3 to 10 (ablation)AutoSelect · AQuAaccuracy: 0.383 -> 0.418Improved
lower initial temperature from 0.1 to 0.02 (ablation)AutoSelect · Trecaccuracy: 0.393 -> 0.334Hurt
raise initial temperature from 0.1 to 0.2 (ablation)AutoSelect · Trecaccuracy: 0.393 -> 0.371Hurt
replay buffer size 50 instead of 10 (ablation)AutoSelect · Trecaccuracy: 0.361 -> 0.393Improved
replay buffer size 100 instead of 50 (ablation)AutoSelect · Epistemic_reasoningaccuracy: 0.601 -> 0.575Hurt
increase validation set size from 100 to 500 (ablation)AutoSelect · Winowhyaccuracy: 0.657 -> 0.635No clear effect
use auto-regressive AutoSelect instead of top-k Qwen3-8B retrievalQwen3-8B · Trecaccuracy: 0.412 -> 0.483Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy