Search papers, labs, and topics across Lattice.
1
0
3
Forget fine-tuning: DeCoVec steers LLMs to better accuracy (+5.5% on average) by simply nudging the decoding logits based on in-context learning.
