Search papers, labs, and topics across Lattice.
1
0
3
Forget hand-engineered reward shaping: ACDC learns robotic manipulation tasks faster and more reliably by adaptively balancing exploration and exploitation with a dynamic contrastive loss.
