{
    "info": {
        "author": "MadryLab",
        "author_email": "ailyas@mit.edu",
        "bugtrack_url": null,
        "classifiers": [
            "Development Status :: 4 - Beta",
            "Intended Audience :: Developers",
            "License :: OSI Approved :: MIT License",
            "Programming Language :: Python :: 3",
            "Programming Language :: Python :: 3.5",
            "Programming Language :: Python :: 3.6",
            "Topic :: Software Development :: Build Tools"
        ],
        "description": "robustness package\n==================\nInstall via ``pip``: ``pip install robustness``\n\nRead the docs: https://robustness.readthedocs.io/en/latest/index.html\n\n``robustness`` is a package we (students in the `MadryLab <http://madry-lab.ml>`_) created\nto make training, evaluating, and exploring neural networks flexible and easy.\nWe use it in almost all of our projects (whether they involve\nadversarial training or not!) and it will be a dependency in many of our\nupcoming code releases. A few projects using the library include:\n\n- `Code for \"Learning Perceptually-Aligned Representations via Adversarial Robustness\" <https://github.com/MadryLab/robust_representations>`_ (https://arxiv.org/abs/1906.00945) \n- `Code for\n  \"Image Synthesis with a Single (Robust) Classifier\" <https://github.com/MadryLab/robustness_applications>`_ (https://arxiv.org/abs/1906.09453)\n\nWe\ndemonstrate how to use the library in a set of walkthroughs and our API\nreference. Functionality provided by the library includes:\n\n- Training and evaluating standard and robust models for a variety of\n  datasets/architectures using a `CLI interface\n  <https://robustness.readthedocs.io/en/latest/example_usage/cli_usage.html>`_. The library also provides support for adding\n  `custom datasets <https://robustness.readthedocs.io/en/latest/example_usage/training_lib_part_2.html#training-on-custom-datasets>`_ and `model architectures <https://robustness.readthedocs.io/en/latest/example_usage/training_lib_part_2.html#training-with-custom-architectures>`_.\n\n.. code-block:: bash\n\n   python -m robustness.main --dataset cifar --data /path/to/cifar \\\n      --adv-train 0 --arch resnet18 --out-dir /logs/checkpoints/dir/\n\n- Performing `input manipulation\n  <https://robustness.readthedocs.io/en/latest/example_usage/input_space_manipulation.html>`_ using robust (or standard)\n  models---this includes making adversarial examples, inverting representations,\n  feature visualization, etc. The library offers a variety of optimization\n  options (e.g. choice between real/estimated gradients, Fourier/pixel basis,\n  custom loss functions etc.), and is easily extendable.\n\n.. code-block:: python\n\n   import torch as ch\n   from robustness.datasets import CIFAR\n   from robustness.model_utils import make_and_restore_model\n\n   ds = CIFAR('/path/to/cifar')\n   model, _ = make_and_restore_model(arch='resnet50', dataset=ds, \n                resume_path='/path/to/model', state_dict_path='model')\n   model.eval()\n   attack_kwargs = {\n      'constraint': 'inf', # L-inf PGD \n      'eps': 0.05, # Epsilon constraint (L-inf norm)\n      'step_size': 0.01, # Learning rate for PGD\n      'iterations': 100, # Number of PGD steps\n      'targeted': True # Targeted attack\n      'custom_loss': None # Use default cross-entropy loss\n   }\n\n   _, test_loader = ds.make_loaders(workers=0, batch_size=10)\n   im, label = next(iter(test_loader))\n   target_label = (label + ch.randint_like(label, high=9)) % 10\n   adv_out, adv_im = model(im, target_label, make_adv, **attack_kwargs)\n\n- Importing ``robustness`` as a package, which allows for easy training of\n  neural networks with support for custom loss functions, logging, data loading,\n  and more! A good introduction can be found in our two-part walkthrough\n  (`Part 1 <https://robustness.readthedocs.io/en/latest/example_usage/training_lib_part_1.html>`_, \n  `Part 2 <https://robustness.readthedocs.io/en/latest/example_usage/training_lib_part_2.html>`_).\n\n.. code-block:: python\n\n   from robustness import model_utils, datasets, train, defaults\n   from robustness.datasets import CIFAR\n\n   # We use cox (http://github.com/MadryLab/cox) to log, store and analyze \n   # results. Read more at https//cox.readthedocs.io.\n   from cox.utils import Parameters\n   import cox.store\n\n   # Hard-coded dataset, architecture, batch size, workers\n   ds = CIFAR('/path/to/cifar')\n   m, _ = model_utils.make_and_restore_model(arch='resnet50', dataset=ds)\n   train_loader, val_loader = ds.make_loaders(batch_size=128, workers=8)\n\n   # Create a cox store for logging\n   out_store = cox.store.Store(OUT_DIR)\n\n   # Hard-coded base parameters\n   train_kwargs = {\n       'out_dir': \"train_out\",\n       'adv_train': 1,\n       'constraint': '2',\n       'eps': 0.5,\n       'attack_lr': 1.5,\n       'attack_steps': 20\n   }\n   train_args = Parameters(train_kwargs)\n\n   # Fill whatever parameters are missing from the defaults\n   train_args = defaults.check_and_fill_args(train_args,\n                           defaults.TRAINING_ARGS, CIFAR)\n   train_args = defaults.check_and_fill_args(train_args,\n                           defaults.PGD_ARGS, CIFAR)\n\n   # Train a model\n   train.train_model(train_args, m, (train_loader, val_loader), store=out_store)\n\nCitation\n--------\nIf you use this library in your research, cite it as\nfollows:\n\n.. code-block:: bibtex\n\n   @misc{robustness,\n      title={Robustness (Python Library)},\n      author={Logan Engstrom and Andrew Ilyas and Shibani Santurkar and Dimitris Tsipras},\n      year={2019},\n      url={https://github.com/MadryLab/robustness}\n   }\n\n*(Have you used the package and found it useful? Let us know!)*.\n\nContributors\n-------------\n- `Andrew Ilyas <https://twitter.com/andrew_ilyas>`_\n- `Logan Engstrom <https://twitter.com/logan_engstrom>`_\n- `Shibani Santurkar <https://twitter.com/ShibaniSan>`_\n- `Dimitris Tsipras <https://twitter.com/tsiprasd>`_\n\n\n\n",
        "description_content_type": "",
        "docs_url": null,
        "download_url": "",
        "downloads": {
            "last_day": -1,
            "last_month": -1,
            "last_week": -1
        },
        "home_page": "",
        "keywords": "logging tools madrylab",
        "license": "MIT",
        "maintainer": "",
        "maintainer_email": "",
        "name": "robustness",
        "package_url": "https://pypi.org/project/robustness/",
        "platform": "",
        "project_url": "https://pypi.org/project/robustness/",
        "project_urls": null,
        "release_url": "https://pypi.org/project/robustness/1.0.post1/",
        "requires_dist": [
            "tqdm",
            "grpcio",
            "psutil",
            "gitpython",
            "py3nvml"
        ],
        "requires_python": "",
        "summary": "Tools for Robustness",
        "version": "1.0.post1"
    },
    "last_serial": 5711087,
    "releases": {
        "1.0": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "3d5359f066a3c63a1ea963ae42d69b18",
                    "sha256": "08706856574c8167a3520d481f1228cafd4848a9c0e469499f7573b0f7422969"
                },
                "downloads": -1,
                "filename": "robustness-1.0-py3-none-any.whl",
                "has_sig": false,
                "md5_digest": "3d5359f066a3c63a1ea963ae42d69b18",
                "packagetype": "bdist_wheel",
                "python_version": "py3",
                "requires_python": null,
                "size": 67245,
                "upload_time": "2019-08-21T10:06:46",
                "url": "https://files.pythonhosted.org/packages/68/ca/3d83745f8d7b797f5f9b1832a26ceb2b45701b68a24f6418d1e7b100d95a/robustness-1.0-py3-none-any.whl"
            },
            {
                "comment_text": "",
                "digests": {
                    "md5": "70c909a301b392ef7bca6a14318a47e8",
                    "sha256": "ffc1cf68bfe3bed267eeb085a8b6142a983a8516c1191ece5cf7d3609de3f387"
                },
                "downloads": -1,
                "filename": "robustness-1.0.tar.gz",
                "has_sig": false,
                "md5_digest": "70c909a301b392ef7bca6a14318a47e8",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 57307,
                "upload_time": "2019-08-21T10:06:49",
                "url": "https://files.pythonhosted.org/packages/e3/ca/9f1a8756161a47b2d8982a2dcf81fb4f83ebfa8e4de86e69f3edac3a28d7/robustness-1.0.tar.gz"
            }
        ],
        "1.0.post1": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "461392605dd6c96d054549372d83ec9e",
                    "sha256": "38c834b4d5b13e023df7262d590e073c6c83725af3d1e2fd30491fcdca3bd4d4"
                },
                "downloads": -1,
                "filename": "robustness-1.0.post1-py3-none-any.whl",
                "has_sig": false,
                "md5_digest": "461392605dd6c96d054549372d83ec9e",
                "packagetype": "bdist_wheel",
                "python_version": "py3",
                "requires_python": null,
                "size": 67541,
                "upload_time": "2019-08-21T18:25:25",
                "url": "https://files.pythonhosted.org/packages/01/31/c9b45a4f6ac909c39e6b01dedbd648258693a3b4d54968838e64a6ca5366/robustness-1.0.post1-py3-none-any.whl"
            },
            {
                "comment_text": "",
                "digests": {
                    "md5": "21bf6865e82dfef9a3d3d8c4dead9347",
                    "sha256": "b09d0b692fbc3850d4d49103ae9ccc7b9e8c03a421a798ea50de51f17e56f3c3"
                },
                "downloads": -1,
                "filename": "robustness-1.0.post1.tar.gz",
                "has_sig": false,
                "md5_digest": "21bf6865e82dfef9a3d3d8c4dead9347",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 57566,
                "upload_time": "2019-08-21T18:25:27",
                "url": "https://files.pythonhosted.org/packages/f5/a5/7565532da7baad8da02d370a0113509ce8bb88ae5916a0e5d7de64ae7279/robustness-1.0.post1.tar.gz"
            }
        ]
    },
    "urls": [
        {
            "comment_text": "",
            "digests": {
                "md5": "461392605dd6c96d054549372d83ec9e",
                "sha256": "38c834b4d5b13e023df7262d590e073c6c83725af3d1e2fd30491fcdca3bd4d4"
            },
            "downloads": -1,
            "filename": "robustness-1.0.post1-py3-none-any.whl",
            "has_sig": false,
            "md5_digest": "461392605dd6c96d054549372d83ec9e",
            "packagetype": "bdist_wheel",
            "python_version": "py3",
            "requires_python": null,
            "size": 67541,
            "upload_time": "2019-08-21T18:25:25",
            "url": "https://files.pythonhosted.org/packages/01/31/c9b45a4f6ac909c39e6b01dedbd648258693a3b4d54968838e64a6ca5366/robustness-1.0.post1-py3-none-any.whl"
        },
        {
            "comment_text": "",
            "digests": {
                "md5": "21bf6865e82dfef9a3d3d8c4dead9347",
                "sha256": "b09d0b692fbc3850d4d49103ae9ccc7b9e8c03a421a798ea50de51f17e56f3c3"
            },
            "downloads": -1,
            "filename": "robustness-1.0.post1.tar.gz",
            "has_sig": false,
            "md5_digest": "21bf6865e82dfef9a3d3d8c4dead9347",
            "packagetype": "sdist",
            "python_version": "source",
            "requires_python": null,
            "size": 57566,
            "upload_time": "2019-08-21T18:25:27",
            "url": "https://files.pythonhosted.org/packages/f5/a5/7565532da7baad8da02d370a0113509ce8bb88ae5916a0e5d7de64ae7279/robustness-1.0.post1.tar.gz"
        }
    ]
}