Generative Flow Networks or GFlowNets is a new paradigm of neural net training, developed at MILA since 2021.

GFlowNets are related to Monte-Carlo Markov chain methods (as they sample from a distribution specified by an energy function), reinforcement learning (as they learn a policy to sample composed objects through a sequence of steps), generative models (as they learn to represent and sample from a distribution) and amortized variational methods (as they can be used to learn to approximate and sample from an otherwise intractable posterior, given a prior and a likelihood). GFlowNet are trained to generate an object x through a sequence of steps with probability proportional to some reward function R(x) (or eE(x) with E(x) denoting the energy function), given at the end of the generative trajectory.[1]

Through generative models and variational inference, GFlowNets are also related to Active Inference.

GFlowNets promise better interpretability and more robust reasoning than the current auto-regressive LLMs[2].

  1. ^

    Pan, L., Malkin, N., Zhang, D., & Bengio, Y. (2023). Better Training of GFlowNets with Local Credit and Incomplete Trajectories (arXiv:2302.01687). arXiv.

  2. ^

    Bengio, Y., & Hu, E. (2023, March 21). Scaling in the service of reasoning & model-based ML. Yoshua Bengio.

Created by Roman Leventov at 9mo