arXiv (All)
open access
Calibration, Not Answer Selection: Distilling Internal Confidence in Reasoning Models
Reinforcement learning with binary correctness rewards trains correctness, not calibrated confidence. The confidence that reasoning models verbalize is systematically overconfident, and the problem is not merely one of scale: verbalized confidence tracks how willing a model is to commit to an answer
Conceptio Open Archive
Analyze & explore related research in Conceptio
Connect this record directly into your research workflow or AI agent. Explore citations across open repositories, query full text, and export structured proof bundles.
This document is indexed with metadata only: full text is not available in the archive for this record. Open the official source ↗
Related documents
- Rewarding Efficient Reasoning Improves Abstention on Underspecified Tasks in Reasoning Models#1034823
- Nudgeability: Reasoning Models Follow Confidence Signals Without Tracking Their Own Competence#1128837
- Learning to Optimize through Solver-Grounded Self-Play#1119278
- Post-Reasoning: Improving the Performance of Non-Thinking Models at No Cost#1084202
- Reasoning Correctness: The Missing Dimension in Evaluating Large Language Models#696814
Record · ID 1118002
Retrieved via Conceptio: every document is proof-bundled with source, license, and retrieval metadata.
Retrieved via Conceptio: every document is proof-bundled with source, license, and retrieval metadata.
Bring this data into your agent: Connect via MCP or API · 20 free credits to start · no card required
Read the Docs → · View CLI source on GitHub · Conceptio Open Knowledge Archive
Read the Docs → · View CLI source on GitHub · Conceptio Open Knowledge Archive