Search papers, labs, and topics across Lattice.
6
0
11
Grounding clinical language models in structured physiological knowledge can boost safety scores by over 21 percentage points, surpassing even state-of-the-art models like GPT-4.
Hierarchical local attention in TextNCA reveals that the arrangement of attention windows can dramatically influence language modeling performance, even more than the model's iterative nature.
MMOE achieves faster convergence and superior generation quality in diffusion transformers by effectively integrating expert routing strategies, challenging the notion that more parameters always lead to better performance.
Emotion classifiers can now provide explanations that are not just post hoc but are grounded in definitional semantics, ensuring transparency and auditability.
A comprehensive taxonomy reveals critical failure modes in LLM reasoning, exposing vulnerabilities that could hinder their deployment in real-world applications.
Standard multimodal fusion can hurt performance in emotion recognition, but this new approach knows when to drop modalities, leading to state-of-the-art results.
