Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
Affiliation:
7
0
8
0
This first study of Web search across four major conversational platforms is presented, combining real-world user interactions with controlled experiments using the same platform's models by their APIs (invitro), finding that Web-search decisions vary substantially across platforms and models.
LLMs struggle to generate robust travel itineraries, with substantial gaps in handling delays and crowd dynamics revealed by UTP-Bench.
Confusion-aware retrieval can boost text classification performance by up to 10 percentage points, even for smaller models.
Travel itineraries generated by TRIPPULSE are not only feasible but also deeply personalized, leveraging user reviews to enhance experiential quality.
MCES reveals that integrating diverse analytical methods can dramatically enhance the reliability of causal inference from observational data, achieving perfect precision in ranking true relationships.
Surprisingly, a trie-guided decoding framework applied to smaller encoder-decoder models like T5 and BART can outperform much larger instruction-tuned models like LLaMA-3 and Phi-3 in in-document query auto-completion.
Forget struggling with ambiguous text descriptions – now you can automatically generate high-fidelity scientific architecture diagrams from natural language, thanks to a new dataset that rivals GPT-4o's diagramming abilities.