Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/danijar/dreamerv3
Mastering Diverse Domains through World Models
https://github.com/danijar/dreamerv3
artificial-intelligence general jax minecraft reinforcement-learning world-models
Last synced: about 1 month ago
JSON representation
Mastering Diverse Domains through World Models
- Host: GitHub
- URL: https://github.com/danijar/dreamerv3
- Owner: danijar
- License: mit
- Created: 2023-01-14T12:08:18.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2024-07-29T22:55:04.000Z (about 1 month ago)
- Last Synced: 2024-07-30T04:17:34.724Z (about 1 month ago)
- Topics: artificial-intelligence, general, jax, minecraft, reinforcement-learning, world-models
- Language: Python
- Homepage: https://danijar.com/dreamerv3
- Size: 25.5 MB
- Stars: 1,182
- Watchers: 26
- Forks: 203
- Open Issues: 34
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
- Awesome-state-space-models - GitHub
README
# Mastering Diverse Domains through World Models
A reimplementation of [DreamerV3][paper], a scalable and general reinforcement
learning algorithm that masters a wide range of applications with fixed
hyperparameters.
If you find this code useful, please reference in your paper:
```
@article{hafner2023dreamerv3,
title={Mastering Diverse Domains through World Models},
author={Hafner, Danijar and Pasukonis, Jurgis and Ba, Jimmy and Lillicrap, Timothy},
journal={arXiv preprint arXiv:2301.04104},
year={2023}
}
```
To learn more:
- [Research paper][paper]
- [Project website][website]
- [Twitter summary][tweet]
## DreamerV3
DreamerV3 learns a world model from experiences and uses it to train an actor
critic policy from imagined trajectories. The world model encodes sensory
inputs into categorical representations and predicts future representations and
rewards given actions.
DreamerV3 masters a wide range of domains with a fixed set of hyperparameters,
outperforming specialized methods. Removing the need for tuning reduces the
amount of expert knowledge and computational resources needed to apply
reinforcement learning.
Due to its robustness, DreamerV3 shows favorable scaling properties. Notably,
using larger models consistently increases not only its final performance but
also its data-efficiency. Increasing the number of gradient steps further
increases data efficiency.
# Instructions
The code has been tested on Linux and Mac and requires Python 3.11+.
## Docker
You can either use the provided `Dockerfile` that contains instructions or
follow the manual instructions below.
## Manual
Install [JAX][jax] and then the other dependencies:
```sh
pip install -U -r embodied/requirements.txt
pip install -U -r dreamerv3/requirements.txt \
-f https://storage.googleapis.com/jax-releases/jax_cuda_releases.html
```
Simple training script:
```sh
python example.py
```
Flexible training script:
```sh
python dreamerv3/main.py \
--logdir ~/logdir/{timestamp} \
--configs crafter \
--run.train_ratio 32
```
To reproduce results, train on the desired task using the corresponding config,
such as `--configs atari --task atari_pong`.
# Tips
- All config options are listed in `configs.yaml` and you can override them
as flags from the command line.
- The `debug` config block reduces the network size, batch size, duration
between logs, and so on for fast debugging (but does not learn a good model).
- By default, the code tries to run on GPU. You can switch to CPU or TPU using
the `--jax.platform cpu` flag.
- You can use multiple config blocks that will override defaults in the
order they are specified, for example `--configs crafter size50m`.
- By default, metrics are printed to the terminal, appended to a JSON lines
file, and written as TensorBoard summaries. Other outputs like WandB can be
enabled in the training script.
- If you get a `Too many leaves for PyTreeDef` error, it means you're
reloading a checkpoint that is not compatible with the current config. This
often happens when reusing an old logdir by accident.
- If you are getting CUDA errors, scroll up because the cause is often just an
error that happened earlier, such as out of memory or incompatible JAX and
CUDA versions. Try `--batch_size 1` to rule out an out of memory error.
- Many environments are included, some of which require installing additional
packages. See the `Dockerfile` for reference.
- When running on custom environments, make sure to specify the observation
keys the agent should be using via the `enc.spaces` and `dec.spaces` regex
patterns.
- To log metrics from environments without showing them to the agent or storing
them in the replay buffer, return them as observation keys with `log_` prefix
and enable logging via the `run.log_keys_...` options.
- To continue stopped training runs, simply run the same command line again and
make sure that the `--logdir` points to the same directory.
# Disclaimer
This repository contains a reimplementation of DreamerV3 based on the open
source DreamerV2 code base. It is unrelated to Google or DeepMind. The
implementation has been tested to reproduce the official results on a range of
environments.
[jax]: https://github.com/google/jax#pip-installation-gpu-cuda
[paper]: https://arxiv.org/pdf/2301.04104v1.pdf
[website]: https://danijar.com/dreamerv3
[tweet]: https://twitter.com/danijarh/status/1613161946223677441
[example]: https://github.com/danijar/dreamerv3/blob/main/example.py