8 papers
EgoSafetyBench: A Diagnostic Egocentric Video Benchmark for Evaluating Embodied VLMs as Runtime Safety Guards
Siddhant Panpatil, Arth Singh, Mijin Koo +3
Vision-language models (VLMs) are increasingly proposed as runtime safety guards for embodied agents in homes and factories. A deployable guard must catch genuinely unsafe situatio…
Constitutional Arms Races in the Public Goods Game: Co-Evolving LLM Constitutions Under Cooperation-Defection Pressure
Ujwal Kumar, Arth Singh, Hershraj Niranjani +5
Frontier LLM agents engage in blackmail, sabotage, and document leaks under goal conflicts in agentic settings, exposing limitations of alignment methods built around single-agent…
Mechanistic Analysis of Alignment Algorithms in Language Models
Aarush Sinha, Ishan Garg, Veeraraju Elluru +2
Post-training alignment algorithms are predominantly evaluated as black boxes, obscuring how they reshape language models' internal computations. We present a systematic mechanisti…
Toward Human-AI Complementarity Across Diverse Tasks
Yuzheng Xu, Annya Dahmani, Matthew D. Blanchard +13
Human-AI complementarity, the idea that combining human and AI judgments can outperform either alone, offers a promising pathway toward robust oversight of advanced AI systems. How…
Re-Mask and Redirect: Exploiting Denoising Irreversibility in Diffusion Language Models
Arth Singh
Safety alignment in diffusion language models (dLLMs) relies on a single load-bearing assumption: that committed tokens are permanent. We show that violating this assumption, by re…
Mechanistically Interpreting Compression in Vision-Language Models
Veeraraju Elluru, Arth Singh, Roberto Aguero +3
Compressed vision-language models (VLMs) are widely used to reduce memory and compute costs, making them a suitable choice for real-world deployment. However, compressing these mod…
EMA Is Not All You Need: Mapping the Boundary Between Structure and Content in Recurrent Context
Arth Singh
What exactly do efficient sequence models gain over simple temporal averaging? We use exponential moving average (EMA) traces, the simplest recurrent context (no gating, no content…
Fluid Reasoning Representations
Dmitrii Kharlapenko, Terry Jingchen Zhang, Arth Singh +4
Frontier large language models increasingly solve complex tasks involving abstract concepts through extended test-time thinking. Yet we lack a mechanistic account of how extended t…