Targeted controls show that tool-invocation scores alone do not establish event grounding and that reminder coverage depends on response history and frequency, and motivate developing and evaluating streaming assistants for timely action, initiative under general user needs, and coverage of entire response sequences.
Yin-Song Yan, Ling Wang, Yuxuan Wang et al.· 0 citations
Autonomous research seeks sustained model improvements through iterative experimentation and feedback. LLM agents show promise in automating machine learning and language-model post-training, but their ability to sustain multimodal improvement remains unclear. We introduce MMPostTrainBench, a benchmark spanning eight t...
Yu-Xin Liu, Yuxuan Wang, Zhen-Xin Lei et al.· 0 citations
Recent advances have enabled unified omni-modal models in understanding audio, vision, and language. However, existing benchmarks, training data, and learning methods largely treat the modalities independently, leaving the capability of audio-visual joint reasoning poorly evaluated and insufficiently elicited. We addre...
Junming Lin, Yuxuan Wang, Zhen-Xin Lei et al.· 0 citations
OmniEchoBench, a unified benchmark for spatial audio-visual perception and audio-vision-language navigation, and a spatially aware omni-modal model, which introduces an FOA spatial encoder alongside a pretrained semantic audio pathway.