Search papers, labs, and topics across Lattice.
Affiliation:
1
0
2
3
Self-correction in LLMs can be dramatically improved by reinforcing step-level reasoning, achieving higher accuracy and reliability in outputs.
