B-Coder: Value-Based Deep Reinforcement Learning for Program Synthesis.

Zishun Yu, Yunzhe Tao, Liyu Chen, Tao Sun, Hongxia Yang

Browse the full ICLR paper archive.