{
    "info": {
        "author": "Anjandeep Singh Sahni",
        "author_email": "sahni.anjandeep@gmail.com",
        "bugtrack_url": null,
        "classifiers": [
            "License :: OSI Approved :: MIT License",
            "Operating System :: OS Independent",
            "Programming Language :: Python :: 3"
        ],
        "description": "==========\nTorchUtils\n==========\n\n.. image:: https://badge.fury.io/py/torchutils.svg\n    :target: https://badge.fury.io/py/torchutils\n\n.. image:: https://travis-ci.org/anjandeepsahni/pytorch_utils.svg?branch=master\n    :target: https://travis-ci.org/anjandeepsahni/pytorch_utils\n\n|\n\n**TorchUtils** is a Python package providing helpful utility functions for your\nPyTorch projects. TorchUtils helps speed up PyTorch development by implementing\ntrivial but useful functionality so that you don't have to.\n\nKey Features\n------------\n\n* Calculate total model parameters.\n* Calculate model FLOPs.\n* Print model summary in Keras style.\n* Save/load checkpoints.\n* Get/set learning rate.\n* Set random seed.\n\nExamples\n--------\n\nModel Summary::\n\n    import torch\n    import torchvision\n    import torchutils as tu\n\n    model = torchvision.models.alexnet()\n    tu.get_model_summary(model, torch.rand((1, 3, 224, 224)))\n\n    # Output\n\n    =========================================================================================\n    Layer                           Kernel             Output          Params           FLOPs\n    =========================================================================================\n    0_features.Conv2d_0         [3, 64, 11, 11]    [1, 64, 55, 55]       23,296    70,470,400\n    1_features.ReLU_1                         -    [1, 64, 55, 55]            0             0\n    2_features.MaxPool2d_2                    -    [1, 64, 27, 27]            0             0\n    3_features.Conv2d_3         [64, 192, 5, 5]   [1, 192, 27, 27]      307,392   224,088,768\n    4_features.ReLU_4                         -   [1, 192, 27, 27]            0             0\n    5_features.MaxPool2d_5                    -   [1, 192, 13, 13]            0             0\n    6_features.Conv2d_6        [192, 384, 3, 3]   [1, 384, 13, 13]      663,936   112,205,184\n    7_features.ReLU_7                         -   [1, 384, 13, 13]            0             0\n    8_features.Conv2d_8        [384, 256, 3, 3]   [1, 256, 13, 13]      884,992   149,563,648\n    9_features.ReLU_9                         -   [1, 256, 13, 13]            0             0\n    10_features.Conv2d_10      [256, 256, 3, 3]   [1, 256, 13, 13]      590,080    99,723,520\n    11_features.ReLU_11                       -   [1, 256, 13, 13]            0             0\n    12_features.MaxPool2d_12                  -     [1, 256, 6, 6]            0             0\n    13_classifier.Dropout_0                   -          [1, 9216]            0             0\n    14_classifier.Linear_1         [9216, 4096]          [1, 4096]   37,752,832    75,493,376\n    15_classifier.ReLU_2                      -          [1, 4096]            0             0\n    16_classifier.Dropout_3                   -          [1, 4096]            0             0\n    17_classifier.Linear_4         [4096, 4096]          [1, 4096]   16,781,312    33,550,336\n    18_classifier.ReLU_5                      -          [1, 4096]            0             0\n    19_classifier.Linear_6         [4096, 1000]          [1, 1000]    4,097,000     8,191,000\n    =========================================================================================\n    Total params: 61,100,840\n    Trainable params: 61,100,840\n    Non-trainable params: 0\n    Total FLOPs: 773,286,232 / 773.29 MFLOPs\n    -----------------------------------------------------------------------------------------\n    Input size (MB): 0.57\n    Forward/backward pass size (MB): 8.31\n    Params size (MB): 233.08\n    Estimated Total Size (MB): 241.96\n    =========================================================================================\n\nLearning Rate::\n\n    import torchvision\n    import torchutils as tu\n    import torch.optim as optim\n\n    model = torchvision.models.alexnet()\n    optimizer = optim.Adam(model.parameters())\n    current_lr = tu.get_lr(optimizer)\n    print('Current learning rate:', current_lr)\n\n    optimizer = tu.set_lr(optimizer, current_lr*0.1)\n    revised_lr = tu.get_lr(optimizer)\n    print('Revised learning rate:', revised_lr)\n\n    # Output\n\n    Current learning rate: 0.001\n    Revised learning rate: 0.0001\n\nCheckpoint::\n\n    import torchvision\n    import torchutils as tu\n    import torch.optim as optim\n\n    model = torchvision.models.alexnet()\n    optimizer = optim.Adam(model.parameters())\n    scheduler = optim.lr_scheduler.ExponentialLR(optimizer, 0.1)\n\n    print('Original learning rate:', tu.get_lr(optimizer))\n\n    # load checkpoint model_20190814-212442_e0_0.7531.pt\n    start_epoch = tu.load_checkpoint(model_path='.',\n                           ckpt_name='model_20190814-212442_e0_0.7531.pt',\n                           model=model, optimizer=optimizer,\n                           scheduler=scheduler)\n\n    print('Checkpoint learning rate:', tu.get_lr(optimizer))\n    print('Start from epoch:', start_epoch)\n\n    # Output\n\n    Original learning rate: 0.001\n    Checkpoint learning rate: 0.1234\n    Start epoch: 1\n\nRequirements\n------------\n\n* Numpy >= 1.16.2\n* PyTorch >= 1.0.0\n\nInstallation\n------------\n\n::\n\n    $ pip install torchutils\n\nDocumentation\n-------------\nAPI documentation is available at: https://anjandeepsahni.github.io/torchutils/\n\nLicense\n-------\nTorchUtils is distributed under the MIT license, see LICENSE.\n\n\n",
        "description_content_type": "",
        "docs_url": null,
        "download_url": "",
        "downloads": {
            "last_day": -1,
            "last_month": -1,
            "last_week": -1
        },
        "home_page": "https://github.com/anjandeepsahni/pytorch_utils",
        "keywords": "machine-learning deep-learning pytorch neuralnetwork",
        "license": "MIT",
        "maintainer": "",
        "maintainer_email": "",
        "name": "torchutils",
        "package_url": "https://pypi.org/project/torchutils/",
        "platform": "",
        "project_url": "https://pypi.org/project/torchutils/",
        "project_urls": {
            "Homepage": "https://github.com/anjandeepsahni/pytorch_utils"
        },
        "release_url": "https://pypi.org/project/torchutils/0.0.2/",
        "requires_dist": null,
        "requires_python": "",
        "summary": "PyTorch utility functions.",
        "version": "0.0.2"
    },
    "last_serial": 5725928,
    "releases": {
        "0.0.1": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "965275ca87e9bdad0d7eb17e5df13509",
                    "sha256": "6447289cc9c630f393f4368536698cb41a3c9e86fbbedae6e4445dc7ecd67995"
                },
                "downloads": -1,
                "filename": "torchutils-0.0.1-py3-none-any.whl",
                "has_sig": false,
                "md5_digest": "965275ca87e9bdad0d7eb17e5df13509",
                "packagetype": "bdist_wheel",
                "python_version": "py3",
                "requires_python": null,
                "size": 4687,
                "upload_time": "2019-07-20T00:47:48",
                "url": "https://files.pythonhosted.org/packages/da/81/e089225a52737e7f3551946d47ed7259e393dfcdf9121f74424caa6d7e3e/torchutils-0.0.1-py3-none-any.whl"
            },
            {
                "comment_text": "",
                "digests": {
                    "md5": "c59384fb8e0243c9b1b8263840eef6aa",
                    "sha256": "3118932cb9f854dc6b0591e9c17a856a7f390530ca1cb94eb51373dc7fe7dbc3"
                },
                "downloads": -1,
                "filename": "torchutils-0.0.1.tar.gz",
                "has_sig": false,
                "md5_digest": "c59384fb8e0243c9b1b8263840eef6aa",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 2587,
                "upload_time": "2019-07-20T00:47:50",
                "url": "https://files.pythonhosted.org/packages/66/30/3f2412f104da2d654a29ba536f758592202c3bb5d780d6730047cb1bea47/torchutils-0.0.1.tar.gz"
            }
        ],
        "0.0.2": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "ba08ab71d6ec36c04b4750f5176216d1",
                    "sha256": "c6bfac7a2a0c7d44982b6c946c4df5f778a99f560fef73ec4cd1980b7294501c"
                },
                "downloads": -1,
                "filename": "torchutils-0.0.2-py3-none-any.whl",
                "has_sig": false,
                "md5_digest": "ba08ab71d6ec36c04b4750f5176216d1",
                "packagetype": "bdist_wheel",
                "python_version": "py3",
                "requires_python": null,
                "size": 14275,
                "upload_time": "2019-08-25T03:38:19",
                "url": "https://files.pythonhosted.org/packages/56/49/e3ca846d4c7862d79200dd9a03fee40a023e1c83f7bb73aa1fde1ec46920/torchutils-0.0.2-py3-none-any.whl"
            },
            {
                "comment_text": "",
                "digests": {
                    "md5": "aef54b6816e4190ca828153f52b7ae97",
                    "sha256": "fb63a2f4fa98ddda5efe267d9cba934dd1121e8d6acf95518fd7f8bf3bc5ca37"
                },
                "downloads": -1,
                "filename": "torchutils-0.0.2.tar.gz",
                "has_sig": false,
                "md5_digest": "aef54b6816e4190ca828153f52b7ae97",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 10532,
                "upload_time": "2019-08-25T03:38:21",
                "url": "https://files.pythonhosted.org/packages/39/18/44bf3e2ca4011aae3487ca1bbad3f580db74e462da28e8fbe91060f13749/torchutils-0.0.2.tar.gz"
            }
        ]
    },
    "urls": [
        {
            "comment_text": "",
            "digests": {
                "md5": "ba08ab71d6ec36c04b4750f5176216d1",
                "sha256": "c6bfac7a2a0c7d44982b6c946c4df5f778a99f560fef73ec4cd1980b7294501c"
            },
            "downloads": -1,
            "filename": "torchutils-0.0.2-py3-none-any.whl",
            "has_sig": false,
            "md5_digest": "ba08ab71d6ec36c04b4750f5176216d1",
            "packagetype": "bdist_wheel",
            "python_version": "py3",
            "requires_python": null,
            "size": 14275,
            "upload_time": "2019-08-25T03:38:19",
            "url": "https://files.pythonhosted.org/packages/56/49/e3ca846d4c7862d79200dd9a03fee40a023e1c83f7bb73aa1fde1ec46920/torchutils-0.0.2-py3-none-any.whl"
        },
        {
            "comment_text": "",
            "digests": {
                "md5": "aef54b6816e4190ca828153f52b7ae97",
                "sha256": "fb63a2f4fa98ddda5efe267d9cba934dd1121e8d6acf95518fd7f8bf3bc5ca37"
            },
            "downloads": -1,
            "filename": "torchutils-0.0.2.tar.gz",
            "has_sig": false,
            "md5_digest": "aef54b6816e4190ca828153f52b7ae97",
            "packagetype": "sdist",
            "python_version": "source",
            "requires_python": null,
            "size": 10532,
            "upload_time": "2019-08-25T03:38:21",
            "url": "https://files.pythonhosted.org/packages/39/18/44bf3e2ca4011aae3487ca1bbad3f580db74e462da28e8fbe91060f13749/torchutils-0.0.2.tar.gz"
        }
    ]
}