Need help with salad?
Click the β€œchat” button below for chat support from the developer who created it, or find similar developers for support.

About the developer

287 Stars 37 Forks Mozilla Public License 2.0 39 Commits 32 Opened issues


A toolbox for domain adaptation and semi-supervised learning. Contributions welcome.

Services available


Need anything else?

Contributors list

πŸ₯— salad

S\ emi-supervised A\ daptive L\ earning A\ cross D\ omains

.. figure:: img/domainshift.png :alt:

is a library to easily setup experiments using the current state-of-the art techniques in domain adaptation. It features several of recent approaches, with the goal of being able to run fair comparisons between algorithms and transfer them to real-world use cases. The toolbox is under active development and will extended when new approaches are published.

Contribute and explore the code on

_. For commonly asked questions, head to our

Check out robusta, our new library for domain adaptation and robustness evaluation on ImageNet scale:

πŸ“Š Benchmarking Results

One of salad's purposes is to constantly track the state of the art of a variety of domain adaptation algorithms. The latest results can be reproduced by the files in the


.. figure:: img/benchmarks.svg :alt:

Code for reproducing these results can be found in the

directory. Usage is outlined below.

πŸ’» Installation

Requirements can be found in

and can be installed via

.. code:: bash

pip install -r requirements.txt

Install the package (recommended) via

.. code:: bash

pip install torch-salad

For the latest development version, install via

.. code:: bash

pip install git+

πŸ“š Using this library

Along with the implementation of domain adaptation routines, this library comprises code to easily set up deep learning experiments in general.

The toolbox currently implements the following techniques (in

) that can be easily run with the provided example script.
  • VADA (

.. code:: bash

    $ python scripts/ --source svhn --target mnist  --vada
  • Domain Adversarial Training (

.. code:: bash

    $ python scripts/ --source svhn --target mnist  --dann  
  • Associative Domain Adaptation (

.. code:: bash

    $ python scripts/ --source svhn --target mnist  --assoc  
  • Deep Correlation Alignment

.. code:: bash

$ python scripts/ --source svhn --target mnist  --coral  
  • Self-Ensembling for Visual Domain Adaptation (

.. code:: bash

   $ python scripts/ --source svhn --target mnist    --teach
  • Adversarial Dropout Regularization (

.. code:: bash

   $ python scripts/ --source svhn --target mnist  --adv  

Examples (already refer to the

subfolder) soon to be added for:
  • Generalizing Across Domains via Cross-Gradient Training (

    __ Example coming soon!
  • DIRT-T (


Implements the following features (in

  • Weights Ensembling using Exponential Moving Averages or Stored Weights
  • WalkerLoss and Visit Loss (
  • Virtual Adversarial Training (

Coming soon:

  • Deep Joint Optimal Transport (
  • Translation based approaches

Quick Start ~~~~~~~~~~~

To get started, the

directory contains several python scripts for both running replication studies on digit benchmarks and studies on a different dataset (toy example: adaptation to noisy images).

.. code:: bash

$ cd scripts
$ python --log ./log --teach --source svhn --target mnist

Refer to the help pages for all options:

.. code::

usage: [-h] [--gpu GPU] [--cpu] [--njobs NJOBS] [--log LOG]
                    [--epochs EPOCHS] [--checkpoint CHECKPOINT]
                    [--learningrate LEARNINGRATE] [--dryrun]
                    [--source {mnist,svhn,usps,synth,synth-small}]
                    [--target {mnist,svhn,usps,synth,synth-small}]
                    [--sourcebatch SOURCEBATCH] [--targetbatch TARGETBATCH]
                    [--seed SEED] [--print] [--null] [--adv] [--vada]
                    [--dann] [--assoc] [--coral] [--teach]

Domain Adaptation Comparision and Reproduction Study

optional arguments: -h, --help show this help message and exit --gpu GPU Specify GPU --cpu Use CPU Training --njobs NJOBS Number of processes per dataloader --log LOG Log directory. Will be created if non-existing --epochs EPOCHS Number of Epochs (Full passes through the unsupervised training set) --checkpoint CHECKPOINT Checkpoint path --learningrate LEARNINGRATE Learning rate for Adam. Defaults to Karpathy's constant ;-) --dryrun Perform a test run, without actually training a network. --source {mnist,svhn,usps,synth,synth-small} Source Dataset. Choose mnist or svhn --target {mnist,svhn,usps,synth,synth-small} Target Dataset. Choose mnist or svhn --sourcebatch SOURCEBATCH Batch size of Source --targetbatch TARGETBATCH Batch size of Target --seed SEED Random Seed --print --null --adv Train a model with Adversarial Domain Regularization --vada Train a model with Virtual Adversarial Domain Adaptation --dann Train a model with Domain Adversarial Training --assoc Train a model with Associative Domain Adaptation --coral Train a model with Deep Correlation Alignment --teach Train a model with Self-Ensembling

Reasons for using solver abstractions ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

The chosen abstraction style organizes experiments into a subclass of


Quickstart: MNIST Experiment ~~~~~~~~~~~~~~~~~~~~~~~~~~~~

As a quick MNIST experiment:

.. code:: python

from salad.solvers import Solver

class MNISTSolver(Solver):

def __init__(self, model, dataset, **kwargs):

    self.model = model
    super().__init__(dataset, **kwargs)

def _init_optims(self, lr = 1e-4, **kwargs):

    opt = torch.optim.Adam(self.model.parameters(), lr = lr)

def _init_losses(self):

For a simple tasks as MNIST, the code is quite long compared to other PyTorch examples


πŸ’‘ Domain Adaptation Problems

Legend: Implemented (βœ“), Under Construction (🚧)

πŸ“· Vision ~~~~~~~~~

  • Digits: MNIST ↔ SVHN ↔ USPS ↔ SYNTH (βœ“)
  • VisDA 2018 Openset and Detection 
    __ (βœ“)
  • Synthetic (GAN) ↔ Real (🚧)
  • CIFAR ↔ STL (🚧)
  • ImageNet to
    __ (🚧)

🎀 Audio ~~~~~~~~

  • Mozilla Common Voice Dataset 
    __ (🚧)

፨ Neuroscience ~~~~~~~~~~~~~~

  • White Noise ↔ Gratings ↔ Natural Images (🚧)
  • Deep Lab Cut Tracking 
    __ (🚧)

πŸ”— References

If you use salad in your publications, please cite

.. code:: bibtex

@misc{schneider2018salad, title={Salad: A Toolbox for Semi-supervised Adaptive Learning Across Domains}, author={Schneider, Steffen and Ecker, Alexander S. and Macke, Jakob H. and Bethge, Matthias}, year={2018}, url={} }

along with the references to the original papers that are implemented here.

Part of the code in this repository is inspired or borrowed from original implementations, especially:


Excellent list of domain adaptation ressources:


Further transfer learning ressources:


πŸ‘€ Contact

Maintained by

Steffen Schneider 
. Work is part of my thesis project at the
Bethge Lab 
. This README is also available as a webpage at 
. We welcome issues and pull requests
to the official github

We use cookies. If you continue to browse the site, you agree to the use of cookies. For more information on our use of cookies please see our Privacy Policy.