When should we prefer Decision Transformers for Offline Reinforcement Learning?

Prajjwal Bhargava, Rohan Chitnis, A. Geramifard, Shagun Sodhani, Amy Zhang
2/12/2026

No official code found. Best implementation: prajjwal1/rl_paradigm (17 stars, MIT, last commit 2024-01-31).

Semantic Scholar

Code Implementations(1)

prajjwal1/rl_paradigmCommunity68%
MIT

Code for ICLR 2024 paper "When should we prefer Decision Transformers for Offline Reinforcement Learning?"

173Shell, PythonMay 4, 20232 years agoMIT

Cite this paper

@article{bhargava2026when,
  title  = {When should we prefer Decision Transformers for Offline Reinforcement Learning?},
  author = {Prajjwal Bhargava and Rohan Chitnis and A. Geramifard and Shagun Sodhani and Amy Zhang},
  year   = {2026},
  url    = {https://api.semanticscholar.org/CorpusID:b2bf9ecff894cc7979ddd656d265e81dc772e370},
  journal = {ICLR 2024 2024}
}

Discussion