TheMachine Press

A daily newspaper for the age of artificial intelligence.

Morning editionPermanent story

benchmarks evals

The Correct Answer Survived Behind the Wrong Score

A two-parameter, label-free correction recovered decisions from hidden states after native sequence scoring collapsed.

Published Updated Story ID: mp-2026-09-01-005
Read the complete editionStory JSON

Summary

A two-parameter, label-free correction recovered decisions from hidden states after native sequence scoring collapsed.

The study distinguishes absent reasoning from an output bottleneck: hidden-state probes decoded correct answers even when native sequence scores failed under structural biases. A minimal additive correction fitted on as few as 25 unlabeled examples recovered 9 to 34 accuracy points for Qwen3.5 models and transferred to OLMo-2-1B and Llama-3.1-8B. Hard-instance and permutation controls supported an instance-specific signal, but the result narrows how some benchmark failures should be interpreted rather than proving latent correctness in every error.

Why it matters

A two-parameter, label-free correction recovered decisions from hidden states after native sequence scoring collapsed.

Limits and context

No additional limitation was separately recorded.

Key claims

  1. A two-parameter, label-free correction recovered decisions from hidden states after native sequence scoring collapsed.

    Evidence: source-2026-09-01-005

Sources

  1. arXiv preprint 2608.31068arXiv · primary research

Corrections

No corrections have been recorded for this story.