Improving GFlowNets with Monte Carlo Tree Search
This work addresses a domain-specific problem for researchers in machine learning focused on generative modeling and sequential decision-making, representing an incremental improvement by building on existing connections between GFlowNets and reinforcement learning.
The paper tackled the problem of improving Generative Flow Networks (GFlowNets) for sampling from distributions over compositional discrete spaces by enhancing their planning capabilities with Monte Carlo Tree Search (MCTS), specifically adapting the MENTS algorithm, resulting in improved sample efficiency during training and generation fidelity of pre-trained models.
Generative Flow Networks (GFlowNets) treat sampling from distributions over compositional discrete spaces as a sequential decision-making problem, training a stochastic policy to construct objects step by step. Recent studies have revealed strong connections between GFlowNets and entropy-regularized reinforcement learning. Building on these insights, we propose to enhance planning capabilities of GFlowNets by applying Monte Carlo Tree Search (MCTS). Specifically, we show how the MENTS algorithm (Xiao et al., 2019) can be adapted for GFlowNets and used during both training and inference. Our experiments demonstrate that this approach improves the sample efficiency of GFlowNet training and the generation fidelity of pre-trained GFlowNet models.