Learning to Stop without Learning to Stop: Self-Supervised Confidence Training Improves Reasoning Efficiency
Self-supervised confidence fine-tuning cuts reasoning tokens by up to 25% without training models to stop early.
Stories per day · 30 days
2 stories, newest first · one card per story
Stories tagged #confidence
Calibration as a First-Class Criterion in LLM Evaluation
Authors argue LLM benchmarks should report calibration alongside accuracy for trustworthy evaluation.