Search papers, labs, and topics across Lattice.
1
0
2
RL fine-tuning might be less about teaching LLMs new tricks and more about activating pre-existing "good vs. bad" representations lurking within them.
