When Linguistic and Internal Confidence Diverge in Large Language Models
Users often ask large language models (LLMs) to report how confident they are, but it is unclear whether such linguistic confidence tracks the model's internal confidence. We study this question across 8 classification tasks, 2 generation tasks and 30 models from three families. For classification, we compare linguistic confidence with logits-based confidence along three axes: association, magnitude agreement and calibration. For generation, we test whether linguistic confidence tracks semantic-entropy-based uncertainty. The axes frequently diverge. Instance-level association is weak on averag
Lineage graph
Paper → model → repo connections mined from source citations (Tier-1 exact match).
Why these links exist
Every edge carries a method, confidence, and the source snippet that justified it — so bad links are debuggable.
- PossiblePossibly related (embedding) · 57%Evaluating J-space entropy as an error predictor across 7 datasets on Qwen3-4B [R] →
- FuzzyOverlapping authors or contributors · 62%DietrichGebert/ponytail →
“Shared author/contributor keys: cheng”
- LinkedLinked via arxiv author · 85%Hefan Zhang →
“When Linguistic and Internal Confidence Diverge in Large Language Models”
- LinkedLinked via arxiv author · 85%Bingquan Zhang →
“When Linguistic and Internal Confidence Diverge in Large Language Models”
- LinkedLinked via arxiv author · 85%Ming-Ming Cheng →
“When Linguistic and Internal Confidence Diverge in Large Language Models”
- LinkedLinked via arxiv author · 85%Saeed Hassanpour →
“When Linguistic and Internal Confidence Diverge in Large Language Models”
- LinkedLinked via arxiv author · 85%Weicheng Ma →
“When Linguistic and Internal Confidence Diverge in Large Language Models”
- LinkedLinked via arxiv author · 85%Soroush Vosoughi →
“When Linguistic and Internal Confidence Diverge in Large Language Models”