LLMs struggle to associate epistemic markers with stable internal confidence levels across distributions, even under model-centric interpretations, while maintaining somewhat consistent marker rankings.
On the probability–quality paradox in language generation
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
years
2026 3verdicts
UNVERDICTED 3representative citing papers
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%.
Reasoning models detect modifications to their chains of thought with only modest accuracy and cannot reliably identify the nature of those modifications.
citing papers explorer
-
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.
-
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%.
-
Can Reasoning Models Detect Changes to their Chains of Thought?
Reasoning models detect modifications to their chains of thought with only modest accuracy and cannot reliably identify the nature of those modifications.