Search papers, labs, and topics across Lattice.
3
0
7
0
Residual accessibility in LLMs reveals a surprising correlation with recovery speed, challenging the notion that optimizing for audit scores leads to genuine knowledge deletion.
MLLMs struggle with visual prompts, showing a 17.8-point accuracy drop when questions are embedded in images, revealing a critical semantic gap in multimodal reasoning.
Achieve real-time video understanding with transparent reasoning: \model{} aligns response timing with visual evidence, offering a breakthrough for online video LLMs.
