TRACED shows correct LLM reasoning as high-progress stable trajectories and hallucinations as low-progress unstable patterns with high curvature fluctuations.
For example, at α= 0.1 (severe hallucination dominance), the AUROC forLlama-3.1-8B decreases by approximately6%compared to the balanced setting
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Beyond Scalars: Evaluating and Understanding LLM Reasoning via Geometric Progress and Stability
TRACED shows correct LLM reasoning as high-progress stable trajectories and hallucinations as low-progress unstable patterns with high curvature fluctuations.