4 papers
NarrativeWorldBench: A Frontier-Saturated Benchmark and a Latent World Model for Long-Horizon Co-Creative Audio Drama
Logan Mann, Abdur Rahman, Mohammad Saifullah +2
Long-form serialized audio drama, with arcs that run for 200 to 800 episodes, is a major creative medium and a setting where frontier large language models (LLMs) fail. We benchmar…
Visuals Lie, Consistency Speaks: Disentangling Spatial Attention from Reliability in Vision-Language Models
Logan Mann, Yi Xia, Ajit Saravanan +6
Multimodal Foundation Models are increasingly used as reasoning agents, making reliability, knowing when a model may hallucinate, critical. A common intuition, which we call the At…
Where Reliability Lives in Vision-Language Models: A Mechanistic Study of Attention, Hidden States, and Causal Circuits
Logan Mann, Ajit Saravanan, Ishan Dave +4
A pervasive intuition holds that vision-language models (VLMs) are most trustworthy when their attention maps look sharp: concentrated attention on the queried region should imply…
Don't Think of the White Bear: Ironic Negation in Transformer Models Under Cognitive Load
Logan Mann, Nayan Saxena, Sarah Tandon +3
Negation instructions such as 'do not mention ' can paradoxically increase the accessibility of in human thought, a phenomenon known as ironic rebound. Large language models…