Decision Transformer casts RL as autoregressive sequence modeling conditioned on desired returns, past states and actions, matching or exceeding offline RL baselines on Atari, Gym and Key-to-Door tasks.
Off-policy deep reinforcement learning without exploration
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.LG 3roles
background 1polarities
background 1representative citing papers
LOM learns a policy from only the highest-expected-return mode of a Gaussian mixture fitted to the behavior policy, yielding gains over full-distribution imitation methods.
citing papers explorer
-
Decision Transformer: Reinforcement Learning via Sequence Modeling
Decision Transformer casts RL as autoregressive sequence modeling conditioned on desired returns, past states and actions, matching or exceeding offline RL baselines on Atari, Gym and Key-to-Door tasks.
-
Learning on One Mode: Addressing Multi-modality in Offline Reinforcement Learning
LOM learns a policy from only the highest-expected-return mode of a Gaussian mixture fitted to the behavior policy, yielding gains over full-distribution imitation methods.
- Lightning OPD: Efficient Post-Training for Large Reasoning Models with Offline On-Policy Distillation