Search papers, labs, and topics across Lattice.
1
0
3
NebulaExp reveals that a meticulously curated dataset and innovative reinforcement learning strategies can boost LLM performance significantly, achieving up to 4.43 points improvement in instruction-following tasks with minimal data.
