Mingyu Wang

Spatial reasoning hallucinations in MLLMs can be drastically reduced by integrating geometric evidence, challenging the effectiveness of traditional mitigation methods.

Fine-grained hallucination diagnosis can dramatically enhance the reliability of multimodal language models by revealing the types of hallucinations they produce and how to correct them.

A 7B parameter model, optimized with multi-task learning and RL, rivals the timeline summarization performance of a 671B parameter model, proving that task-specific fine-tuning can dramatically shrink model size without sacrificing quality.