{
    "info": {
        "author": "Omri Bar",
        "author_email": "baromri@gmail.com",
        "bugtrack_url": null,
        "classifiers": [
            "License :: OSI Approved :: MIT License",
            "Operating System :: OS Independent",
            "Programming Language :: Python :: 3"
        ],
        "description": "# pytorch-checkpoint\n\n[![PyPI version](https://badge.fury.io/py/pytorchcheckpoint.svg)](https://badge.fury.io/py/pytorchcheckpoint)\n\n\nThis package supports saving and loading PyTorch training checkpoints. It is useful when trying the resume model training from a previous step, and can become handy when working with spot instances or when trying to reproduce results.\n\nA model is saved not only with its weights, as one might do for later inference, but the entire state of the model, including the optimizer state and parameters.\n\nIn addition, it allows saving metrics and other values generated while training, such as accuracy and loss values. This makes it possible to recreate the learning curves from past values and continue to update them as training proceed.\n\n\n------------------\n\n\n## Prerequisites\nDeveloped with **Python 3.7.3**, but should be compatible with previous Python version.\n```\npip install torch==1.1.0 torchvision==0.3.0\n```\n\n## Installation\n```pip install pytorchcheckpoint```\n\n## Usage\n```python\nfrom pytorchcheckpoint.checkpoint import CheckpointHandler\ncheckpoint_handler = CheckpointHandler()\n```\n\n#### Storing a general value\n```python\ncheckpoint_handler.store_var(var_name='num_of_classes', value=1000)\n```\n#### Reading a general value\n```python\nnum_of_classes = checkpoint_handler.get_var(var_name='num_of_classes')\n```\n\n#### Storing values and metrics for each epoch/iteration. For example, the loss value: \n```python\ncheckpoint_handler.store_running_var(var_name='loss', iteration=0, value=1.0)\ncheckpoint_handler.store_running_var(var_name='loss', iteration=1, value=0.9)\ncheckpoint_handler.store_running_var(var_name='loss', iteration=2, value=0.8)\n```\n\n#### Reading stored values for epoch/iteration\n```python\nloss = checkpoint_handler.get_running_var(var_name='loss', iteration=0)\n```\n\n#### Storing values and metrics per set: train/valid/test for each epoch/iteration. For example, the top1 value of the train and valid sets: \n```python\ncheckpoint_handler.store_running_var_with_header(header='train', var_name='top1', iteration=0, value=80)\ncheckpoint_handler.store_running_var_with_header(header='train', var_name='top1', iteration=1, value=85)\ncheckpoint_handler.store_running_var_with_header(header='train', var_name='top1', iteration=2, value=90)\ncheckpoint_handler.store_running_var_with_header(header='train', var_name='top1', iteration=3, value=91)\n\ncheckpoint_handler.store_running_var_with_header(header='valid', var_name='top1', iteration=0, value=70)\ncheckpoint_handler.store_running_var_with_header(header='valid', var_name='top1', iteration=1, value=75)\ncheckpoint_handler.store_running_var_with_header(header='valid', var_name='top1', iteration=2, value=80)\ncheckpoint_handler.store_running_var_with_header(header='valid', var_name='top1', iteration=3, value=85)\n```\n\n#### Reading stored values per set: train/valid/test for epoch/iteration\n```python\nloss = checkpoint_handler.get_running_var_with_header(header='train', var_name='loss', iteration=0)\n```\n\n#### Save checkpoint:\n```python\nimport torchvision.models as models\nfrom torch import optim\ncheckpoint_handler.store_running_var(var_name='loss', iteration=0, value=1.0)\nmodel = models.resnet18()\noptimizer = optim.SGD(model.parameters(), lr=0.001, momentum=0.9)\npath2save = '/tmp'\ncheckpoint_path = checkpoint_handler.generate_checkpoint_path(path2save=path2save)\ncheckpoint_handler.save_checkpoint(checkpoint_path=checkpoint_path, iteration=25, model=model, optimizer=optimizer)\n```\n\n#### Load checkpoint:\n```python\ncheckpoint_path = '<checkpoint_path>'\ncheckpoint_handler = checkpoint_handler.load_checkpoint(checkpoint_path)\n```\n\n",
        "description_content_type": "text/markdown",
        "docs_url": null,
        "download_url": "",
        "downloads": {
            "last_day": -1,
            "last_month": -1,
            "last_week": -1
        },
        "home_page": "https://github.com/bomri/pytorch-checkpoint.git",
        "keywords": "",
        "license": "",
        "maintainer": "",
        "maintainer_email": "",
        "name": "pytorchcheckpoint",
        "package_url": "https://pypi.org/project/pytorchcheckpoint/",
        "platform": "",
        "project_url": "https://pypi.org/project/pytorchcheckpoint/",
        "project_urls": {
            "Homepage": "https://github.com/bomri/pytorch-checkpoint.git"
        },
        "release_url": "https://pypi.org/project/pytorchcheckpoint/0.0.5/",
        "requires_dist": null,
        "requires_python": "",
        "summary": "Support PyTorch checkpoints",
        "version": "0.0.5"
    },
    "last_serial": 5336365,
    "releases": {
        "0.0.1": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "083a3562e7b4a0d6a45b0c1264e54a52",
                    "sha256": "82de1360518a9955a4d68ffb9cd09e1bf19c4a1c96ae1305b10de787e248752e"
                },
                "downloads": -1,
                "filename": "pytorchcheckpoint-0.0.1-py3-none-any.whl",
                "has_sig": false,
                "md5_digest": "083a3562e7b4a0d6a45b0c1264e54a52",
                "packagetype": "bdist_wheel",
                "python_version": "py3",
                "requires_python": null,
                "size": 2965,
                "upload_time": "2019-05-23T14:47:21",
                "url": "https://files.pythonhosted.org/packages/c0/25/a561a47e87e3b4d501d9a97a99c23352b921eae3ca04c48369692f214fd2/pytorchcheckpoint-0.0.1-py3-none-any.whl"
            },
            {
                "comment_text": "",
                "digests": {
                    "md5": "dc19d73d8e84efb115da71012343a6f9",
                    "sha256": "9e0d4f2b94ee169873cf27a9300e6241e0a9486838e3bb3d07578d9997bc38bf"
                },
                "downloads": -1,
                "filename": "pytorchcheckpoint-0.0.1.tar.gz",
                "has_sig": false,
                "md5_digest": "dc19d73d8e84efb115da71012343a6f9",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 2274,
                "upload_time": "2019-05-23T14:47:24",
                "url": "https://files.pythonhosted.org/packages/48/40/5dd443af4d12ffb18b4c2bd1b512ebb88b63ef1b8ff4a24b82e2676a43d9/pytorchcheckpoint-0.0.1.tar.gz"
            }
        ],
        "0.0.2": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "5ec9f317123797ac3507cff98aa9d8e3",
                    "sha256": "40d9e153400194f2a710e2bfcb427af529b7ff5f7999970e1f635246b2df5fc6"
                },
                "downloads": -1,
                "filename": "pytorchcheckpoint-0.0.2-py3-none-any.whl",
                "has_sig": false,
                "md5_digest": "5ec9f317123797ac3507cff98aa9d8e3",
                "packagetype": "bdist_wheel",
                "python_version": "py3",
                "requires_python": null,
                "size": 3874,
                "upload_time": "2019-05-23T19:45:51",
                "url": "https://files.pythonhosted.org/packages/85/c9/f9479b03e46fd2690e288d380952c0aae94a8b7c0a5c49d0674cd83108ab/pytorchcheckpoint-0.0.2-py3-none-any.whl"
            },
            {
                "comment_text": "",
                "digests": {
                    "md5": "44f2fe5c0d41c8d842b480c63ec1e453",
                    "sha256": "f0d331e8d6e843f2345f7b93f0c7d877fe6652178b8762ec3c0d21ad319ee876"
                },
                "downloads": -1,
                "filename": "pytorchcheckpoint-0.0.2.tar.gz",
                "has_sig": false,
                "md5_digest": "44f2fe5c0d41c8d842b480c63ec1e453",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 3160,
                "upload_time": "2019-05-23T19:45:53",
                "url": "https://files.pythonhosted.org/packages/44/14/52db4867a9164b7a6f5cba371f345b0117b8fefd561e1d87cf6bdd4a931c/pytorchcheckpoint-0.0.2.tar.gz"
            }
        ],
        "0.0.3": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "5c72b591e55f0587bc99b2ecaf41d23c",
                    "sha256": "612f87f22339dc9f34472589887c5664c3013920eb7998610f81e0ac03966475"
                },
                "downloads": -1,
                "filename": "pytorchcheckpoint-0.0.3-py3-none-any.whl",
                "has_sig": false,
                "md5_digest": "5c72b591e55f0587bc99b2ecaf41d23c",
                "packagetype": "bdist_wheel",
                "python_version": "py3",
                "requires_python": null,
                "size": 4004,
                "upload_time": "2019-05-26T13:13:31",
                "url": "https://files.pythonhosted.org/packages/00/2e/985c048817b403e42b260ae2d3072e4fa4e3c5e1185a753be653b700268b/pytorchcheckpoint-0.0.3-py3-none-any.whl"
            },
            {
                "comment_text": "",
                "digests": {
                    "md5": "288ca2b2d9cdc9756aa5eb8d19e41bb3",
                    "sha256": "906cde99289e1d90b92693fc1a578706475760b9026c253b6e4390656bbf37ba"
                },
                "downloads": -1,
                "filename": "pytorchcheckpoint-0.0.3.tar.gz",
                "has_sig": false,
                "md5_digest": "288ca2b2d9cdc9756aa5eb8d19e41bb3",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 3470,
                "upload_time": "2019-05-26T13:13:33",
                "url": "https://files.pythonhosted.org/packages/bd/3c/339109730ed8be5084689945595c6c83009fc952d3991fdbfa5c2f6b6c51/pytorchcheckpoint-0.0.3.tar.gz"
            }
        ],
        "0.0.4": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "8776b2278d91e69d47c677d2a237e61c",
                    "sha256": "8814a79c9755657a9efeb5daf9077622d963a8cc70af0a4ca3d80fda2833ae30"
                },
                "downloads": -1,
                "filename": "pytorchcheckpoint-0.0.4-py3-none-any.whl",
                "has_sig": false,
                "md5_digest": "8776b2278d91e69d47c677d2a237e61c",
                "packagetype": "bdist_wheel",
                "python_version": "py3",
                "requires_python": null,
                "size": 4164,
                "upload_time": "2019-05-29T08:10:01",
                "url": "https://files.pythonhosted.org/packages/18/44/8bccd319ef59752c59dd2ee1138d48a7013db28e92240e5145d86c3a2671/pytorchcheckpoint-0.0.4-py3-none-any.whl"
            },
            {
                "comment_text": "",
                "digests": {
                    "md5": "6457b4fa7e600ef425920cac3ef460d0",
                    "sha256": "cce94cfc386db3b0ee24a9fc8395b95c140f63c852fbd5c7a9d9709b785e16ae"
                },
                "downloads": -1,
                "filename": "pytorchcheckpoint-0.0.4.tar.gz",
                "has_sig": false,
                "md5_digest": "6457b4fa7e600ef425920cac3ef460d0",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 3743,
                "upload_time": "2019-05-29T08:10:02",
                "url": "https://files.pythonhosted.org/packages/39/15/c660d1d338fb6143951e0724fa1db3963f7a8dc45f13f70828f64102b8e9/pytorchcheckpoint-0.0.4.tar.gz"
            }
        ],
        "0.0.5": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "443bd657b9cede13e92513aed641bd6f",
                    "sha256": "193836b93c61550557f7ebc48bd7fccdd8b74e9268f1f9d8c24abd3bf5afb2ab"
                },
                "downloads": -1,
                "filename": "pytorchcheckpoint-0.0.5-py3-none-any.whl",
                "has_sig": false,
                "md5_digest": "443bd657b9cede13e92513aed641bd6f",
                "packagetype": "bdist_wheel",
                "python_version": "py3",
                "requires_python": null,
                "size": 4246,
                "upload_time": "2019-05-30T08:55:33",
                "url": "https://files.pythonhosted.org/packages/1e/27/d5620c35329772919fdf7e8d30f99f2cc24adc915c2d7d7f4e1dee6ba2af/pytorchcheckpoint-0.0.5-py3-none-any.whl"
            },
            {
                "comment_text": "",
                "digests": {
                    "md5": "4429276cfa30ac48cb4ec5e429bd5782",
                    "sha256": "9f5f8736940475d06513bfbd5b9e7724cdaa647a79cb3dc431f3d5d06606064f"
                },
                "downloads": -1,
                "filename": "pytorchcheckpoint-0.0.5.tar.gz",
                "has_sig": false,
                "md5_digest": "4429276cfa30ac48cb4ec5e429bd5782",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 3821,
                "upload_time": "2019-05-30T08:55:35",
                "url": "https://files.pythonhosted.org/packages/b5/dc/5aa0f499fc0630f36779a847246c0bfcddd6585bddbe0f1107025c558817/pytorchcheckpoint-0.0.5.tar.gz"
            }
        ]
    },
    "urls": [
        {
            "comment_text": "",
            "digests": {
                "md5": "443bd657b9cede13e92513aed641bd6f",
                "sha256": "193836b93c61550557f7ebc48bd7fccdd8b74e9268f1f9d8c24abd3bf5afb2ab"
            },
            "downloads": -1,
            "filename": "pytorchcheckpoint-0.0.5-py3-none-any.whl",
            "has_sig": false,
            "md5_digest": "443bd657b9cede13e92513aed641bd6f",
            "packagetype": "bdist_wheel",
            "python_version": "py3",
            "requires_python": null,
            "size": 4246,
            "upload_time": "2019-05-30T08:55:33",
            "url": "https://files.pythonhosted.org/packages/1e/27/d5620c35329772919fdf7e8d30f99f2cc24adc915c2d7d7f4e1dee6ba2af/pytorchcheckpoint-0.0.5-py3-none-any.whl"
        },
        {
            "comment_text": "",
            "digests": {
                "md5": "4429276cfa30ac48cb4ec5e429bd5782",
                "sha256": "9f5f8736940475d06513bfbd5b9e7724cdaa647a79cb3dc431f3d5d06606064f"
            },
            "downloads": -1,
            "filename": "pytorchcheckpoint-0.0.5.tar.gz",
            "has_sig": false,
            "md5_digest": "4429276cfa30ac48cb4ec5e429bd5782",
            "packagetype": "sdist",
            "python_version": "source",
            "requires_python": null,
            "size": 3821,
            "upload_time": "2019-05-30T08:55:35",
            "url": "https://files.pythonhosted.org/packages/b5/dc/5aa0f499fc0630f36779a847246c0bfcddd6585bddbe0f1107025c558817/pytorchcheckpoint-0.0.5.tar.gz"
        }
    ]
}