Search papers, labs, and topics across Lattice.
2
0
6
7
Vision-language models can now achieve remarkable out-of-distribution accuracy by effectively leveraging self-reflection to correct errors in multimodal reasoning.
Off-the-shelf LLMs can get a 25% boost on long-context reasoning tasks simply by dynamically emphasizing relevant tokens during decoding, without any further training.
