Search papers, labs, and topics across Lattice.
1
7
3
11
Forget fine-tuning: this evolutionary approach lets you adapt LLMs to new tasks with just 200 samples and no gradients, outperforming standard methods by up to 54.8%.
