Search papers, labs, and topics across Lattice.
3
0
5
11
Exploiting local parameter perturbations of a teacher model unlocks complementary supervision signals, boosting LLM mathematical reasoning accuracy beyond standard on-policy self-distillation.
Leveraging its own failure history, DiagEvo transforms self-play by dynamically guiding question generation to address specific reasoning weaknesses, leading to unprecedented accuracy gains.
Self-improvement in LLMs can be achieved without external supervision by leveraging their own evaluative capabilities, leading to substantial performance gains across diverse benchmarks.
