Search papers, labs, and topics across Lattice.
T_{s}:1.2) 41.6 61.1 20.0 33.3 77.4 90.6 50.0 77.5 22.4 43.4 38.1 60.5 GRPO (Ts:1.
1
1
3
2
Forget fixed temperature schedules: TAMPO learns to adapt temperature on-the-fly, boosting LLM reinforcement learning performance on mathematical reasoning tasks.
