AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

Automatically Finding Reward Model Biases

Zifan Wang, Iván Arcuschin, Arthur Conmy

ICML 2026 · Large language models › Language, safety and interpretability · approach: Statistical frameworks for LLM evaluation · 0 citations (OpenAlex)

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
evolutionary multi-step search (depth 5, branching 4) instead of best-of-N (depth 1) (ablation)bias discovery pipeline · Skywork-V2-8B, 6 topicsdiversity-adjusted bias strength / Pareto plotImproved
depth 3, branching 8 instead of depth 1 best-of-N (ablation)bias discovery pipeline · Skywork-V2-8B, 6 topicsdiversity-adjusted bias strengthImproved
evolve triple-space attribute from extra spacing (step 2) to triple spaces (step 3)triple-space lineage · training batchRM bias strength: 0.91 -> 0.41; judge 0.22 -> 0.03Mixed
mutate typographical-error attribute to extra spacing across steps 0 to 2triple-space lineage · training batchRM bias strength: -0.40 -> 0.91Improved
broad prompt mixture vs narrow how-to datasetSkywork-V2-8B format bias · format biaseslist preferenceMixed
lower signal-to-noise in injected affirmation biashypothesis generation · toy recall setuprecall rateNo clear effect
attribute presence rate in sampled responses neither too high nor too lowhypothesis generation · toy recall setuprecall rateImproved
Bonferroni-corrected validation with three rewriterspipeline · Skywork-V2-8B test setsignificant biases: 17 candidates -> 10 significant for both RM and judgeMixed

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