Almanac
paper

From Correctness to Utility: Gain-Based Prefix Evaluation for LLM Reasoning

paperactiveprovisionalfrom-correctness-to-utility-gain-based-prefix-evaluation-for-llm-reasoning-9b206b83·1 events·first seen 9d ago

Aliases: From Correctness to Utility: Gain-Based Prefix Evaluation for LLM Reasoning

Co-occurring entities

More like this (12)

Recent events (1)

6arXiv · cs.CL·9d ago·source ↗

Prefix Utility Model (PUM) trains process reward models on outcome-grounded prefix gain rather than step correctness

A new arXiv preprint proposes replacing local step-correctness signals in process reward models with 'prefix gain' — the improvement in solve-rate induced by conditioning a student model on a given reasoning prefix. The authors train a Prefix Utility Model (PUM) using a pairwise ranking objective and evaluate it across Best-of-N selection, beam search, and RL on mathematical reasoning tasks. PUM shows particular strength when candidate pools are large, search budgets are high, or rule-based rewards are sparse. Code, data, and models are released publicly.