Search papers, labs, and topics across Lattice.
3
0
6
0
Current multimodal LLMs still struggle to integrate information and reason critically when assessed on real scientific papers, despite progress on isolated tasks.
Ramen achieves robust test-time adaptation of VLMs in mixed-domain scenarios by selecting the right samples to adapt to, sidestepping the common pitfall of performance degradation when faced with diverse and inconsistent test data.
Federated learning with LLMs for recommendation can be improved by sharing directional update components and learning personalized integration weights, outperforming naive aggregation.
