A verifiable empirical win rate reward combined with gradient masking enables RL training of a 7B model to reach betting-market calibration on NFL win probabilities using only outcome data.
Linguistic calibration of language models
6 Pith papers cite this work. Polarity classification is still indexing.
6
Pith papers citing it
citation-role summary
background 1
citation-polarity summary
verdicts
UNVERDICTED 6roles
background 1polarities
background 1representative citing papers
LLMs struggle to associate epistemic markers with stable internal confidence levels across distributions, even under model-centric interpretations, while maintaining somewhat consistent marker rankings.
Task calibration aligns LLM distributions in latent task spaces to make MBR decoding provably optimal and improve generation quality.
RLMF uses quality of model self-judgments to refine RL rankings and select training data, achieving SOTA faithful calibration while preserving accuracy and outperforming standard RL by up to 63%.
VLAConf is a one-class discriminative method that estimates step-wise task-success confidence for VLA models via anomaly scoring on frozen representations plus step-conditioned modeling, shown to be more efficient than ensemble or probability baselines on LIBERO and real robots.
SEPs approximate semantic entropy from single-generation hidden states to enable cheap and robust hallucination detection in LLMs.
citing papers explorer
-
Verifiable Rewards for Calibrated Probabilistic Forecasting
A verifiable empirical win rate reward combined with gradient masking enables RL training of a 7B model to reach betting-market calibration on NFL win probabilities using only outcome data.
-
Can LLMs Use Linguistic Uncertainty Markers to Reliably Reflect Intrinsic Confidence?
LLMs struggle to associate epistemic markers with stable internal confidence levels across distributions, even under model-centric interpretations, while maintaining somewhat consistent marker rankings.
-
Task-Aware Calibration: Provably Optimal Decoding in LLMs
Task calibration aligns LLM distributions in latent task spaces to make MBR decoding provably optimal and improve generation quality.
-
Reinforcement Learning with Metacognitive Feedback Elicits Faithful Uncertainty Expression in LLMs
RLMF uses quality of model self-judgments to refine RL rankings and select training data, achieving SOTA faithful calibration while preserving accuracy and outperforming standard RL by up to 63%.
-
VLAConf: Calibrated Task-Success Confidence for Vision-Language-Action Models
VLAConf is a one-class discriminative method that estimates step-wise task-success confidence for VLA models via anomaly scoring on frozen representations plus step-conditioned modeling, shown to be more efficient than ensemble or probability baselines on LIBERO and real robots.
-
Semantic Entropy Probes: Robust and Cheap Hallucination Detection in LLMs
SEPs approximate semantic entropy from single-generation hidden states to enable cheap and robust hallucination detection in LLMs.