Search papers, labs, and topics across Lattice.
2
0
4
11
AHR achieves state-of-the-art performance in text classification with a fraction of the parameters, revolutionizing how LLMs can adapt to limited data.
Mamba models have hidden performance just waiting to be unlocked: simply scaling the activations of identified "bottleneck" subspaces boosts performance by over 8% across diverse tasks.
