Search papers, labs, and topics across Lattice.
1
0
3
Active-GRPO not only outperforms existing methods in molecular optimization but also redefines how models can adaptively balance imitation and self-discovery during training.
