arXiv preprint arXiv:2510.01581 , year=

Think Right: Learning to Mitigate Under-Over Thinking via Adaptive, Attentive Compression , author= · arXiv 2510.01581

1 Pith paper cite this work. Polarity classification is still indexing.

1 Pith paper citing it

representative citing papers

cs.LG · 2026-07-09 · conditional · novelty 7.0

Thinking-mode VLMs collapse answer-token entropy, but thinking-chain entropy and length serve as robust, zero-cost hallucination predictors.

Showing 1 of 1 citing paper.

When Thinking Hurts: Epistemic Signals in the Reasoning Chains of Visual Language Models cs.LG · 2026-07-09 · conditional · none · ref 11
Thinking-mode VLMs collapse answer-token entropy, but thinking-chain entropy and length serve as robust, zero-cost hallucination predictors.