Edward Beeching
added checkpoints and readme file
7978894
|
raw
history blame
537 Bytes
metadata
tags:
  - deep-reinforcement-learning
  - reinforcement-learning

Find here pretrained model weights for the [Decision Transformer] (https://github.com/kzl/decision-transformer). Weights are available for 4 Atari games: Breakout, Pong, Qbert and Seaquest. Found in the checkpoints directory. We share models trained for one seed (123), whereas the paper contained weights for 3 random seeds.

Usage

conda env create -f conda_env.yml

Then, you can use the model like this: