Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
Affiliation:
1
0
2
LLMs struggle with self-awareness, making systematic errors on simple counterfactuals, but reinforcement learning can boost their self-modeling skills.