Concept生成流网络 / Generative Flow Networks
GFlowNets
A class of generative models that learn a stochastic policy such that the probability of generating an object is proportional to its reward, enabling the generation of high-quality samples while maintaining diversity.
Timeline (last 90 days)
Oct 30
GFlowNets sample proportionally rather than maximizing reward, so even when the out-of-sample accuracy of the reward proxy is low, the model maintains broad exploration and is robust to an imperfect reward proxy.
Unverified50%