Search papers, labs, and topics across Lattice.
1
0
3
2
Forget fine-tuning: unlock surprisingly strong performance by stitching together frozen LLMs with lightweight, trainable projections that route information between models.
