AI Paper Atlas

Research map / Large language models / Agents, code and math

When Greedy Wins: Emergent Exploitation Bias in Meta-Bandit LLM Training

Sanxing Chen, Xiaoyin Chen, Yukun Huang, Roy Xie, Bhuwan Dhingra

ICLR 2026 · Large language models › Agents, code and math · approach: Agent instruction-following benchmarks

Read the paperCheck your idea against 137,217 papers

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
RL or SFT training on bandit tasksQwen2.5-7B-Instruct · Gaussian5_Var1_MeanN0 (in-distribution)AvgReward@300: 0.79 -> 1.01 to 1.05Improved
use algorithmic (UCB-imitation) reward instead of raw bandit reward (ablation)Qwen2.5-7B RL · Gaussian5_Var1_MeanN0 (in-distribution)BestArmFreq@300 (%): 79.8 -> 85.7Improved
use strategic regret-shaped reward instead of raw reward (ablation)Qwen2.5-7B RL · Gaussian5_Var1_MeanN0 (in-distribution)BestArmFreq@300 (%): 79.8 -> 81.1Improved
use strategic reward in low-variance Bernoulli training (ablation)Qwen2.5-7B RL · Bernoulli5_UniformregretNo clear effect
use strategic reward on 3B model instead of raw reward (ablation)Qwen2.5-3B RL · Gaussian5_Var1_MeanN0 (in-distribution)BestArmFreq@300 (%): 49.1 -> 56.9Improved
RL on task reward without teacherQwen2.5-3B RL-OG · Gaussian5_Var1_MeanN0 (in-distribution)BestArmFreq@300 (%): 32.7 -> 49.1Mixed
add teacher guidance (RL-ALG or SFT) for 3B model (ablation)Qwen2.5-3B · Gaussian5_Var1_MeanN0 (in-distribution)BestArmFreq@300 (%): 49.1 (RL-OG) -> 71.2 (RL-ALG), 80.8 (SFT)Improved
train with RL/SFT (vs pretrained)Qwen2.5-7B · Gaussian5_Var1_MeanN0 (in-distribution)SuffixFail@150 (%): 0.0 -> 4.7 to 9.4Hurt
imitate UCB with RL-ALGQwen2.5-7B vs UCB teacher · Gaussian5_Var1_MeanN0 (in-distribution)BestArmFreq@300 (%): 80.6 -> 85.7Improved
SFT trained on Bernoulli, evaluated on GaussianQwen2.5 SFT · Gaussian5_Var1_MeanN0 (OOD)worst-case regretHurt
use RL-ALG instead of SFT for OOD transfer (ablation)Qwen2.5-7B · cross-distributionworst-case regretImproved
train in low variance, test on harder high-variance MeanU environmentslearned policies · Gaussian5_VarX_MeanUregretHurt

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