Skip to main content

No project description provided

Project description

Unit-tests Documentation Benchmarks codecov Twitter Follow Python version GitHub license pypi version pypi nightly version Downloads Downloads Discord Shield


Documentation | TensorDict | Features | Examples, tutorials and demos | Citation | Installation | Asking a question | Contributing

TorchRL is an open-source Reinforcement Learning (RL) library for PyTorch.

It provides pytorch and python-first, low and high level abstractions for RL that are intended to be efficient, modular, documented and properly tested. The code is aimed at supporting research in RL. Most of it is written in python in a highly modular way, such that researchers can easily swap components, transform them or write new ones with little effort.

This repo attempts to align with the existing pytorch ecosystem libraries in that it has a dataset pillar (torchrl/envs), transforms, models, data utilities (e.g. collectors and containers), etc. TorchRL aims at having as few dependencies as possible (python standard library, numpy and pytorch). Common environment libraries (e.g. OpenAI gym) are only optional.

On the low-level end, torchrl comes with a set of highly re-usable functionals for cost functions, returns and data processing.

TorchRL aims at (1) a high modularity and (2) good runtime performance. Read the full paper for a more curated description of the library.

Getting started

Check our Getting Started tutorials for quickly ramp up with the basic features of the library!

Documentation and knowledge base

The TorchRL documentation can be found here. It contains tutorials and the API reference.

TorchRL also provides a RL knowledge base to help you debug your code, or simply learn the basics of RL. Check it out here.

We have some introductory videos for you to get to know the library better, check them out:

Writing simplified and portable RL codebase with TensorDict

RL algorithms are very heterogeneous, and it can be hard to recycle a codebase across settings (e.g. from online to offline, from state-based to pixel-based learning). TorchRL solves this problem through TensorDict, a convenient data structure(1) that can be used to streamline one's RL codebase. With this tool, one can write a complete PPO training script in less than 100 lines of code!

import torch
from tensordict.nn import TensorDictModule
from tensordict.nn.distributions import NormalParamExtractor
from torch import nn

from torchrl.collectors import SyncDataCollector
from import TensorDictReplayBuffer, \
    LazyTensorStorage, SamplerWithoutReplacement
from torchrl.envs.libs.gym import GymEnv
from torchrl.modules import ProbabilisticActor, ValueOperator, TanhNormal
from torchrl.objectives import ClipPPOLoss
from torchrl.objectives.value import GAE

env = GymEnv("Pendulum-v1")
model = TensorDictModule(
        nn.Linear(3, 128), nn.Tanh(),
        nn.Linear(128, 128), nn.Tanh(),
        nn.Linear(128, 128), nn.Tanh(),
        nn.Linear(128, 2),
    out_keys=["loc", "scale"]
critic = ValueOperator(
        nn.Linear(3, 128), nn.Tanh(),
        nn.Linear(128, 128), nn.Tanh(),
        nn.Linear(128, 128), nn.Tanh(),
        nn.Linear(128, 1),
actor = ProbabilisticActor(
    in_keys=["loc", "scale"],
    distribution_kwargs={"min": -1.0, "max": 1.0},
buffer = TensorDictReplayBuffer(
collector = SyncDataCollector(
loss_fn = ClipPPOLoss(actor, critic, gamma=0.99)
optim = torch.optim.Adam(loss_fn.parameters(), lr=2e-4)
adv_fn = GAE(value_network=critic, gamma=0.99, lmbda=0.95, average_gae=True)
for data in collector:  # collect data
    for epoch in range(10):
        adv_fn(data)  # compute advantage
        for i in range(20):  # consume data
            sample = buffer.sample(50)  # mini-batch
            loss_vals = loss_fn(sample)
            loss_val = sum(
                value for key, value in loss_vals.items() if
    print(f"avg reward: {data['next', 'reward'].mean().item(): 4.4f}")

Here is an example of how the environment API relies on tensordict to carry data from one function to another during a rollout execution: Alt Text

TensorDict makes it easy to re-use pieces of code across environments, models and algorithms.


For instance, here's how to code a rollout in TorchRL:

- obs, done = env.reset()
+ tensordict = env.reset()
policy = SafeModule(
    in_keys=["observation_pixels", "observation_vector"],
out = []
for i in range(n_steps):
-     action, log_prob = policy(obs)
-     next_obs, reward, done, info = env.step(action)
-     out.append((obs, next_obs, action, log_prob, reward, done))
-     obs = next_obs
+     tensordict = policy(tensordict)
+     tensordict = env.step(tensordict)
+     out.append(tensordict)
+     tensordict = step_mdp(tensordict)  # renames next_observation_* keys to observation_*
- obs, next_obs, action, log_prob, reward, done = [torch.stack(vals, 0) for vals in zip(*out)]
+ out = torch.stack(out, 0)  # TensorDict supports multiple tensor operations

Using this, TorchRL abstracts away the input / output signatures of the modules, env, collectors, replay buffers and losses of the library, allowing all primitives to be easily recycled across settings.


Here's another example of an off-policy training loop in TorchRL (assuming that a data collector, a replay buffer, a loss and an optimizer have been instantiated):

- for i, (obs, next_obs, action, hidden_state, reward, done) in enumerate(collector):
+ for i, tensordict in enumerate(collector):
-     replay_buffer.add((obs, next_obs, action, log_prob, reward, done))
+     replay_buffer.add(tensordict)
    for j in range(num_optim_steps):
-         obs, next_obs, action, hidden_state, reward, done = replay_buffer.sample(batch_size)
-         loss = loss_fn(obs, next_obs, action, hidden_state, reward, done)
+         tensordict = replay_buffer.sample(batch_size)
+         loss = loss_fn(tensordict)

This training loop can be re-used across algorithms as it makes a minimal number of assumptions about the structure of the data.

TensorDict supports multiple tensor operations on its device and shape (the shape of TensorDict, or its batch size, is the common arbitrary N first dimensions of all its contained tensors):

# stack and cat
tensordict = torch.stack(list_of_tensordicts, 0)
tensordict =, 0)
# reshape
tensordict = tensordict.view(-1)
tensordict = tensordict.permute(0, 2, 1)
tensordict = tensordict.unsqueeze(-1)
tensordict = tensordict.squeeze(-1)
# indexing
tensordict = tensordict[:2]
tensordict[:, 2] = sub_tensordict
# device and memory location

TensorDict comes with a dedicated tensordict.nn module that contains everything you might need to write your model with it. And it is functorch and torch.compile compatible!

transformer_model = nn.Transformer(nhead=16, num_encoder_layers=12)
+ td_module = SafeModule(transformer_model, in_keys=["src", "tgt"], out_keys=["out"])
src = torch.rand((10, 32, 512))
tgt = torch.rand((20, 32, 512))
+ tensordict = TensorDict({"src": src, "tgt": tgt}, batch_size=[20, 32])
- out = transformer_model(src, tgt)
+ td_module(tensordict)
+ out = tensordict["out"]

The TensorDictSequential class allows to branch sequences of nn.Module instances in a highly modular way. For instance, here is an implementation of a transformer using the encoder and decoder blocks:

encoder_module = TransformerEncoder(...)
encoder = TensorDictSequential(encoder_module, in_keys=["src", "src_mask"], out_keys=["memory"])
decoder_module = TransformerDecoder(...)
decoder = TensorDictModule(decoder_module, in_keys=["tgt", "memory"], out_keys=["output"])
transformer = TensorDictSequential(encoder, decoder)
assert transformer.in_keys == ["src", "src_mask", "tgt"]
assert transformer.out_keys == ["memory", "output"]

TensorDictSequential allows to isolate subgraphs by querying a set of desired input / output keys:

transformer.select_subsequence(out_keys=["memory"])  # returns the encoder
transformer.select_subsequence(in_keys=["tgt", "memory"])  # returns the decoder

Check TensorDict tutorials to learn more!


  • A common interface for environments which supports common libraries (OpenAI gym, deepmind control lab, etc.)(1) and state-less execution (e.g. Model-based environments). The batched environments containers allow parallel execution(2). A common PyTorch-first class of tensor-specification class is also provided. TorchRL's environments API is simple but stringent and specific. Check the documentation and tutorial to learn more!

    env_make = lambda: GymEnv("Pendulum-v1", from_pixels=True)
    env_parallel = ParallelEnv(4, env_make)  # creates 4 envs in parallel
    tensordict = env_parallel.rollout(max_steps=20, policy=None)  # random rollout (no policy given)
    assert tensordict.shape == [4, 20]  # 4 envs, 20 steps rollout
    env_parallel.action_spec.is_in(tensordict["action"])  # spec check returns True
  • multiprocess and distributed data collectors(2) that work synchronously or asynchronously. Through the use of TensorDict, TorchRL's training loops are made very similar to regular training loops in supervised learning (although the "dataloader" -- read data collector -- is modified on-the-fly):

    env_make = lambda: GymEnv("Pendulum-v1", from_pixels=True)
    collector = MultiaSyncDataCollector(
        [env_make, env_make],
        devices=["cuda:0", "cuda:0"],
    for i, tensordict_data in enumerate(collector):
        loss = loss_module(tensordict_data)

    Check our distributed collector examples to learn more about ultra-fast data collection with TorchRL.

  • efficient(2) and generic(1) replay buffers with modularized storage:

    storage = LazyMemmapStorage(  # memory-mapped (physical) storage
    buffer = TensorDictPrioritizedReplayBuffer(
        collate_fn=lambda x: x,
        pin_memory=device != torch.device("cpu"),
        prefetch=10,  # multi-threaded sampling

    Replay buffers are also offered as wrappers around common datasets for offline RL:

    from import SamplerWithoutReplacement
    from import D4RLExperienceReplay
    data = D4RLExperienceReplay(
    for sample in data:  # or alternatively sample = data.sample()
  • cross-library environment transforms(1), executed on device and in a vectorized fashion(2), which process and prepare the data coming out of the environments to be used by the agent:

    env_make = lambda: GymEnv("Pendulum-v1", from_pixels=True)
    env_base = ParallelEnv(4, env_make, device="cuda:0")  # creates 4 envs in parallel
    env = TransformedEnv(
            ObservationNorm(loc=0.5, scale=1.0)),  # executes the transforms once and on device
    tensordict = env.reset()
    assert tensordict.device == torch.device("cuda:0")

    Other transforms include: reward scaling (RewardScaling), shape operations (concatenation of tensors, unsqueezing etc.), concatenation of successive operations (CatFrames), resizing (Resize) and many more.

    Unlike other libraries, the transforms are stacked as a list (and not wrapped in each other), which makes it easy to add and remove them at will:

    env.insert_transform(0, NoopResetEnv())  # inserts the NoopResetEnv transform at the index 0

    Nevertheless, transforms can access and execute operations on the parent environment:

    transform = env.transform[1]  # gathers the second transform of the list
    parent_env = transform.parent  # returns the base environment of the second transform, i.e. the base env + the first transform
  • various tools for distributed learning (e.g. memory mapped tensors)(2);

  • various architectures and models (e.g. actor-critic)(1):

    # create an nn.Module
    common_module = ConvNet(
        num_cells=[32, 64, 64],
        kernel_sizes=[8, 4, 3],
        strides=[4, 2, 1],
    # Wrap it in a SafeModule, indicating what key to read in and where to
    # write out the output
    common_module = SafeModule(
    # Wrap the policy module in NormalParamsWrapper, such that the output
    # tensor is split in loc and scale, and scale is mapped onto a positive space
    policy_module = SafeModule(
            MLP(num_cells=[64, 64], out_features=32, activation=nn.ELU)
        out_keys=["loc", "scale"],
    # Use a SafeProbabilisticTensorDictSequential to combine the SafeModule with a
    # SafeProbabilisticModule, indicating how to build the
    # torch.distribution.Distribution object and what to do with it
    policy_module = SafeProbabilisticTensorDictSequential(  # stochastic policy
            in_keys=["loc", "scale"],
    value_module = MLP(
        num_cells=[64, 64],
    # Wrap the policy and value funciton in a common module
    actor_value = ActorValueOperator(common_module, policy_module, value_module)
    # standalone policy from this
    standalone_policy = actor_value.get_policy_operator()
  • exploration wrappers and modules to easily swap between exploration and exploitation(1):

    policy_explore = EGreedyWrapper(policy)
    with set_exploration_type(ExplorationType.RANDOM):
        tensordict = policy_explore(tensordict)  # will use eps-greedy
    with set_exploration_type(ExplorationType.MODE):
        tensordict = policy_explore(tensordict)  # will not use eps-greedy
  • A series of efficient loss modules and highly vectorized functional return and advantage computation.


    Loss modules

    from torchrl.objectives import DQNLoss
    loss_module = DQNLoss(value_network=value_network, gamma=0.99)
    tensordict = replay_buffer.sample(batch_size)
    loss = loss_module(tensordict)

    Advantage computation

    from torchrl.objectives.value.functional import vec_td_lambda_return_estimate
    advantage = vec_td_lambda_return_estimate(gamma, lmbda, next_state_value, reward, done, terminated)
  • a generic trainer class(1) that executes the aforementioned training loop. Through a hooking mechanism, it also supports any logging or data transformation operation at any given time.

  • various recipes to build models that correspond to the environment being deployed.

If you feel a feature is missing from the library, please submit an issue! If you would like to contribute to new features, check our call for contributions and our contribution page.

Examples, tutorials and demos

A series of examples are provided with an illustrative purpose:

and many more to come!

Check the examples markdown directory for more details about handling the various configuration settings.

We also provide tutorials and demos that give a sense of what the library can do.


If you're using TorchRL, please refer to this BibTeX entry to cite this work:

      title={TorchRL: A data-driven decision-making library for PyTorch}, 
      author={Albert Bou and Matteo Bettini and Sebastian Dittert and Vikash Kumar and Shagun Sodhani and Xiaomeng Yang and Gianni De Fabritiis and Vincent Moens},


Create a conda environment where the packages will be installed.

conda create --name torch_rl python=3.9
conda activate torch_rl


Depending on the use of functorch that you want to make, you may want to install the latest (nightly) PyTorch release or the latest stable version of PyTorch. See here for a detailed list of commands, including pip3 or other special installation instructions.


You can install the latest stable release by using

pip3 install torchrl

This should work on linux, Windows 10 and OsX (Intel or Silicon chips). On certain Windows machines (Windows 11), one should install the library locally (see below).

The nightly build can be installed via

pip install torchrl-nightly

which we currently only ship for Linux and OsX (Intel) machines. Importantly, the nightly builds require the nightly builds of PyTorch too.

To install extra dependencies, call

pip3 install "torchrl[atari,dm_control,gym_continuous,rendering,tests,utils,marl,checkpointing]"

or a subset of these.

One may also desire to install the library locally. Three main reasons can motivate this:

  • the nightly/stable release isn't available for one's platform (eg, Windows 11, nightlies for Apple Silicon etc.);
  • contributing to the code;
  • install torchrl with a previous version of PyTorch (note that this should also be doable via a regular install followed by a downgrade to a previous pytorch version -- but the C++ binaries will not be available.)

To install the library locally, start by cloning the repo:

git clone

Go to the directory where you have cloned the torchrl repo and install it (after installing ninja)

cd /path/to/torchrl/
pip install ninja -U
python develop

(unfortunately, pip install -e . will not work).

On M1 machines, this should work out-of-the-box with the nightly build of PyTorch. If the generation of this artifact in MacOs M1 doesn't work correctly or in the execution the message (mach-o file, but is an incompatible architecture (have 'x86_64', need 'arm64e')) appears, then try

ARCHFLAGS="-arch arm64" python develop

To run a quick sanity check, leave that directory (e.g. by executing cd ~/) and try to import the library.

python -c "import torchrl"

This should not return any warning or error.

Optional dependencies

The following libraries can be installed depending on the usage one wants to make of torchrl:

# diverse
pip3 install tqdm tensorboard "hydra-core>=1.1" hydra-submitit-launcher

# rendering
pip3 install moviepy

# deepmind control suite
pip3 install dm_control

# gym, atari games
pip3 install "gym[atari]" "gym[accept-rom-license]" pygame

# tests
pip3 install pytest pyyaml pytest-instafail

# tensorboard
pip3 install tensorboard

# wandb
pip3 install wandb


If a ModuleNotFoundError: No module named ‘torchrl._torchrl errors occurs (or a warning indicating that the C++ binaries could not be loaded), it means that the C++ extensions were not installed or not found.

  • One common reason might be that you are trying to import torchrl from within the git repo location. The following code snippet should return an error if torchrl has not been installed in develop mode:
    cd ~/path/to/rl/repo
    python -c 'from torchrl.envs.libs.gym import GymEnv'
    If this is the case, consider executing torchrl from another location.
  • If you're not importing torchrl from within its repo location, it could be caused by a problem during the local installation. Check the log after the python develop. One common cause is a g++/C++ version discrepancy and/or a problem with the ninja library.
  • If the problem persists, feel free to open an issue on the topic in the repo, we'll make our best to help!
  • On MacOs, we recommend installing XCode first. With Apple Silicon M1 chips, make sure you are using the arm64-built python (e.g. here). Running the following lines of code
    should display
    OS: macOS *** (arm64)
    and not
    OS: macOS **** (x86_64)

Versioning issues can cause error message of the type undefined symbol and such. For these, refer to the versioning issues document for a complete explanation and proposed workarounds.

Asking a question

If you spot a bug in the library, please raise an issue in this repo.

If you have a more generic question regarding RL in PyTorch, post it on the PyTorch forum.


Internal collaborations to torchrl are welcome! Feel free to fork, submit issues and PRs. You can checkout the detailed contribution guide here. As mentioned above, a list of open contributions can be found in here.

Contributors are recommended to install pre-commit hooks (using pre-commit install). pre-commit will check for linting related issues when the code is committed locally. You can disable th check by appending -n to your commit command: git commit -m <commit message> -n


This library is released as a PyTorch beta feature. BC-breaking changes are likely to happen but they will be introduced with a deprecation warranty after a few release cycles.


TorchRL is licensed under the MIT License. See LICENSE for details.

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distributions

No source distribution files available for this release.See tutorial on generating distribution archives.

Built Distributions

torchrl-0.4.0-cp312-cp312-macosx_11_0_arm64.whl (1.2 MB view hashes)

Uploaded CPython 3.12 macOS 11.0+ ARM64

torchrl-0.4.0-cp311-cp311-win_amd64.whl (893.0 kB view hashes)

Uploaded CPython 3.11 Windows x86-64

torchrl-0.4.0-cp311-cp311-manylinux1_x86_64.whl (6.0 MB view hashes)

Uploaded CPython 3.11

torchrl-0.4.0-cp311-cp311-macosx_11_0_arm64.whl (1.2 MB view hashes)

Uploaded CPython 3.11 macOS 11.0+ ARM64

torchrl-0.4.0-cp310-cp310-win_amd64.whl (894.9 kB view hashes)

Uploaded CPython 3.10 Windows x86-64

torchrl-0.4.0-cp310-cp310-manylinux1_x86_64.whl (5.9 MB view hashes)

Uploaded CPython 3.10

torchrl-0.4.0-cp310-cp310-macosx_11_0_arm64.whl (1.2 MB view hashes)

Uploaded CPython 3.10 macOS 11.0+ ARM64

torchrl-0.4.0-cp39-cp39-win_amd64.whl (892.1 kB view hashes)

Uploaded CPython 3.9 Windows x86-64

torchrl-0.4.0-cp39-cp39-manylinux1_x86_64.whl (5.9 MB view hashes)

Uploaded CPython 3.9

torchrl-0.4.0-cp39-cp39-macosx_11_0_arm64.whl (1.2 MB view hashes)

Uploaded CPython 3.9 macOS 11.0+ ARM64

torchrl-0.4.0-cp38-cp38-win_amd64.whl (894.9 kB view hashes)

Uploaded CPython 3.8 Windows x86-64

torchrl-0.4.0-cp38-cp38-manylinux1_x86_64.whl (6.0 MB view hashes)

Uploaded CPython 3.8

torchrl-0.4.0-cp38-cp38-macosx_11_0_arm64.whl (1.3 MB view hashes)

Uploaded CPython 3.8 macOS 11.0+ ARM64

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page