Proximal Policy Optimization for Amortized Discrete Sampling
For researchers using GFlowNets for discrete sampling, this work provides a more efficient training algorithm, though it is an incremental adaptation of existing RL methods.
This paper derives and applies proximal policy optimization (PPO) to the Generative Flow Network (GFlowNet) framework for discrete sampling, achieving improved convergence speed and data efficiency on benchmarks including synthetic energies and molecular graph generation.
This paper explores policy gradient algorithms for training stochastic policies to sample from structured discrete probability distributions under the Generative Flow Network (GFlowNet) framework. Building on extensive theoretical connections between GFlowNets and entropy-regularized reinforcement learning, we derive equivalents of standard policy gradient algorithms for training GFlowNets, as well as experimentally explore their various methodological aspects, including baseline training and advantage estimation. Most importantly, our work is the first to derive and successfully apply proximal policy optimization to GFlowNets, showing its improved convergence speed and data efficiency compared to standard GFlowNet training objectives on benchmarks ranging from synthetic energies to molecular graph generation.