arXiv AI
Taming CoT Obfuscation in VLMs: From Mechanistic Evidence to Activation-Level Enforcement
Read the original on arXiv AI →The Flow has not summarised this story yet — read it at arXiv AI.
Related stories
arXiv AI
Sep 23
Taming CoT Obfuscation in VLMs: From Mechanistic Evidence to Activation Enforcement
The paper investigates how reinforcement learning can unintentionally obscure the chain‑of‑thought (CoT) reasoning in vision‑language models, making their internal reasoning less traceable. By analyzing activation patterns, the authors show that template‑associated activations become less distinguishable during RL and that targeted interventions can mitigate this effect. They introduce TAME, a method that uses sparse autoencoders to suppress these problematic activations while still encouraging accurate behavior, achieving significant gains in CoT monitorability across multiple datasets and model families.
arXiv Machine Learning
Oct 5
From Patching to Pruning Visual Computation in Vision Language Models
arXiv:2610.03389v1 Announce Type: cross Abstract: Vision language models (VLMs) incur substantial inference cost because every visual token is processed by the attention and MLP projections of every...
arXiv Computer Vision
Oct 16
Perception or Reasoning? Diagnosing and Tackling Reinforcement Learning Bottlenecks in VLMs
The paper introduces Perception and Reasoning Attribution (PARA), a method that quantifies how much of a vision‑language model’s errors can be fixed by better perception versus stronger reasoning. PARA reveals that reinforcement learning (RL) benefits more from improving reasoning than perception, and motivates a new group‑relative attention distillation objective called GRAD. GRAD directly guides weaker rollouts toward the visual attention patterns of the best rollouts, consistently boosting performance across multiple model scales and benchmarks.
arXiv Computer Vision
Oct 12
In-Context Learning Can Help Vision-Language Models Overcome Training Priors
The paper shows that controlled visual in‑context learning can help vision‑language models (VLMs) overcome errors caused by training priors. By providing demonstrations, the models focus more on visual evidence, yielding significant accuracy gains on counterfactual examples while leaving prior‑aligned performance largely unchanged. This effect is observed across five VLM families and five vision‑centric tasks, indicating that many VLMs possess latent visual grounding abilities that standard tests may miss.
arXiv AI
Jun 4
Bad Seeing or Bad Thinking? Rewarding Perception for Multimodal Reasoning
arXiv:2605. 14054v2 Announce Type: replace Abstract: Achieving robust perception-reasoning synergy is a central goal for advanced Vision-Language Models (VLMs).
arXiv AI
Jul 15
Visual Access Boundaries in Vision-Language Model Reasoning
arXiv:2607. 12815v1 Announce Type: new Abstract: Chain-of-Thought (CoT) prompting is widely used as a test-time scaling strategy for Vision-Language Models (VLMs), but it remains unclear what is extended when VLMs generate longer reasoning traces.