// ♞ Reinforcement Learning
Decision Transformer
RL as sequence modelling: condition on a target return and predict the next action.
2021 Intermediate extended
IN PLAIN WORDS
Write a trajectory as interleaved (return-to-go, state, action) tokens and train a GPT-style causal Transformer with plain supervised learning. At test time, prompt with the return you want and it acts accordingly. No value functions or policy gradients, yet it matched dedicated offline-RL methods of the time.
Used in: A research baseline for offline RL and sequence-model control
TIMELINE · prerequisites → this → what builds on it
01Original paper
Decision Transformer: Reinforcement Learning via Sequence Modeling
Lili Chen, Kevin Lu, Aravind Rajeswaran, et al. · 2021
arXiv ↗ ✓ Verified · arXiv API
02Where it sits
Learn the left column first; the right column builds on this one.
Then you can learn
- Nothing in the atlas builds on it yet
Related:Conservative Q-Learning (CQL, Offline RL)World Models
Left: prerequisites · right: builds on this · dashed: related. Full map →
03Learning resources
Official docs, explainers, courses and reference code. All links checked.
04Where the course teaches it
For AI / data
- Verification: verified, checked 2026-10-22.
- JSON: /api/learn/algorithms.json (id:
decision-transformer)