infrastructure
Model Disagreement Can Route the Hard Cases
Weak-to-strong escalation works without ground-truth labels by comparing multiple local answers.
Summary
Weak-to-strong escalation works without ground-truth labels by comparing multiple local answers.
A small model often cannot know when it is wrong, and collecting correctness labels for open-ended agent work is expensive. This routing method compares multiple weak-model responses and estimates disagreement against a stronger model as a proxy for when escalation is worthwhile. Across agentic and single-turn tasks, it balanced cost and accuracy without ground-truth supervision, sometimes outperforming routers trained directly on correctness labels. Its approximation improves as the stronger model better tracks the true answer distribution.
Why it matters
Weak-to-strong escalation works without ground-truth labels by comparing multiple local answers.
Limits and context
- A small model often cannot know when it is wrong, and collecting correctness labels for open-ended agent work is expensive.
Key claims
Weak-to-strong escalation works without ground-truth labels by comparing multiple local answers.
Qualification: A small model often cannot know when it is wrong, and collecting correctness labels for open-ended agent work is expensive.
Evidence: source-2026-10-18-018
Sources
- arXiv preprint 2610.22245, original v1arXiv · primary research
Corrections
No corrections have been recorded for this story.