kzl/decision-transformer
PyTorch implementation of Decision Transformer, a reinforcement learning model that reformulates RL as autoregressive sequence modeling.
Not currently ranked — collecting fresh signals.
star history
The repository provides the official implementation of Decision Transformer, a 2021 paper that treats reinforcement learning as a sequence modeling problem. It uses a transformer architecture to predict actions based on past states, returns, and actions, enabling offline RL without explicit value estimation. The codebase includes scripts for Atari games and OpenAI Gym environments to reproduce the paper’s experimental results.
Frequently asked
- What is kzl/decision-transformer?
- PyTorch implementation of Decision Transformer, a reinforcement learning model that reformulates RL as autoregressive sequence modeling.
- Is decision-transformer open source?
- Yes — kzl/decision-transformer is open source, released under the MIT license.
- What language is decision-transformer written in?
- kzl/decision-transformer is primarily written in Python.
- How popular is decision-transformer?
- kzl/decision-transformer has 2.8k stars on GitHub.
- Where can I find decision-transformer?
- kzl/decision-transformer is on GitHub at https://github.com/kzl/decision-transformer.