Skip to main content
Search

# Temporal Difference Flows

1 related articles

Subscribe to updates
  1. arXiv

    Goals no longer have to come from real trajectories: TD-Aug imagines them with geometric horizon models, lifting average success on stitching tasks from 31 to 79

    Offline goal-conditioned behavior cloning generalizes poorly on tasks that require stitching, because its training goals come only from future states within a single trajectory. TD-Aug instead samples the mixture policy's successor measure from a geometric horizon model and trains the policy on those imagined long-horizon states as extra goals. On OGBench pointmaze stitching tasks, average success rises from 31 to 79, above offline goal-conditioned RL baselines GCIQL, QRL, CRL and HIQL.

You’re all caught up