Search papers, labs, and topics across Lattice.
2
0
2
Large reasoning models don't need longer chains of thought to perform better—projecting drifting hidden states back into an efficient latent cluster cuts token generation by up to 53% while boosting task accuracy by up to 12.1%.
A single generalist model can achieve state-of-the-art performance in Referring Expression Comprehension while generalizing across diverse datasets without fine-tuning.
