Search papers, labs, and topics across Lattice.
1
0
2
Transforming off-policy tokens into on-policy ones could be the key to unlocking more robust and efficient alignment for large language models.
