Controlling Exploration-Exploitation in GFlowNets via Markov Chain Perspectives

TL;DR

This paper links GFlowNet objectives to Markov chain properties, revealing constraints on exploration-exploitation, and introduces $eta$-GFNs to better control this trade-off, improving mode discovery in generative tasks.

Contribution

It establishes a theoretical connection between GFlowNets and Markov chain reversibility, and proposes $eta$-GFNs for tunable exploration-exploitation control.

Findings

$eta$-GFNs outperform previous GFlowNets in mode discovery.

Up to 10x increase in discovered modes across benchmarks.

Theoretical framework linking GFlowNets to Markov chain properties.

Abstract

Generative Flow Network (GFlowNet) objectives implicitly fix an equal mixing of forward and backward policies, potentially constraining the exploration-exploitation trade-off during training. By further exploring the link between GFlowNets and Markov chains, we establish an equivalence between GFlowNet objectives and Markov chain reversibility, thereby revealing the origin of such constraints, and provide a framework for adapting Markov chain properties to GFlowNets. Building on these theoretical findings, we propose $\alpha$-GFNs, which generalize the mixing via a tunable parameter $\alpha$. This generalization enables direct control over exploration-exploitation dynamics to enhance mode discovery capabilities, while ensuring convergence to unique flows. Across various benchmarks, including Set, Bit Sequence, and Molecule Generation, $\alpha$-GFN objectives consistently outperform previous GFlowNet objectives, achieving up to a $10 \times$ increase in the number of discovered modes.