{
    "info": {
        "author": "Dan Shiebler",
        "author_email": "danshiebler@gmail.com",
        "bugtrack_url": null,
        "classifiers": [
            "License :: OSI Approved :: MIT License",
            "Operating System :: OS Independent",
            "Programming Language :: Python :: 3"
        ],
        "description": "# repcomp\n\n`repcomp` (short for representation comparison) is a package for comparing trained embedding models. You can use it to compare Deep Neural Networks, Matrix Factorization models, Graph Embeddings, Word Embeddings, etc.\n\n`repcomp` supports the following embedding comparison approaches:\n\n* Nearest Neighbors: Fetch the nearest neighbor set of each entity according to embedding distances, and compare model A's neighbor sets to model B's neighbor sets.\n* Canonical Correlation: Treat embedding components as observations of random variables and compute the canonical correlations between model A and model B. \n* Unit Match: Form a unit-to-unit matching between model A's embedding components and model B's embedding components and measure the correlations of the matched units.\n\nA simple example comparing random embeddings:\n\n```python\n  from repcomp.comparison import CCAComparison\n  import numpy as np\n\n  # Generate random embedding matrices\n  num_samples = 100\n  num_components = 10\n  embedding_1 = np.random.random((num_samples, num_components))\n  embedding_2 = embedding_1 + 0.5 * np.random.random((num_samples, num_components))\n\n  # Run the comparison\n  comparator = CCAComparison()\n  sim = comparator.run_comparison(embedding_1, embedding_2)\n  print(\"The canonical correlation similarity is {}\".format(sim[\"similarity\"]))\n```\n\nA more involved example comparing word embeddings:\n\n```python\n  import gensim.downloader as api\n  import numpy as np\n  from repcomp.comparison import NeighborsComparison\n\n  # Load word vectors from gensim\n  glove_wiki_50 = api.load(\"glove-wiki-gigaword-50\")\n  glove_twitter_50 = api.load(\"glove-twitter-50\")\n\n  # Build the embedding matrices over the shared vocabularies\n  shared_vocab = set(glove_wiki_50.vocab.keys()).intersection(\n    set(glove_twitter_50.vocab.keys()))\n  glove_wiki_50_vectors = np.vstack([glove_wiki_50.get_vector(word) for word in shared_vocab])\n  glove_twitter_50_vectors = np.vstack([glove_twitter_50.get_vector(word) for word in shared_vocab])\n\n  # Run the comparison\n  comparator = NeighborsComparison()\n  print(\"The neighbors similarity between glove-wiki-gigaword-50 and glove-twitter-50 is {}\".format(\n    comparator.run_comparison(glove_wiki_50_vectors, glove_twitter_50_vectors)[\"similarity\"]))\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/pypa/sampleproject",
        "keywords": "",
        "license": "",
        "maintainer": "",
        "maintainer_email": "",
        "name": "repcomp",
        "package_url": "https://pypi.org/project/repcomp/",
        "platform": "",
        "project_url": "https://pypi.org/project/repcomp/",
        "project_urls": {
            "Homepage": "https://github.com/pypa/sampleproject"
        },
        "release_url": "https://pypi.org/project/repcomp/0.1/",
        "requires_dist": [
            "pandas (>=0.20.3)",
            "gensim (>=0.13.3)",
            "keras (>=2.0.1)",
            "annoy (>=1.12.0)",
            "mock (>=2.0.0)",
            "tqdm (>=4.19.5)",
            "scikit-learn (>=0.19.1)",
            "pytest (>=3.4.1)"
        ],
        "requires_python": "",
        "summary": "A package for comparing trained embedding models.",
        "version": "0.1"
    },
    "last_serial": 4388591,
    "releases": {
        "0.1": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "f301b48f4aa2e11e60bb2c86144c1aa8",
                    "sha256": "52184ecd2ffbdbef4587040a71d69ae1301caf8ab450ea357f6f96cfd4415808"
                },
                "downloads": -1,
                "filename": "repcomp-0.1-py3-none-any.whl",
                "has_sig": false,
                "md5_digest": "f301b48f4aa2e11e60bb2c86144c1aa8",
                "packagetype": "bdist_wheel",
                "python_version": "py3",
                "requires_python": null,
                "size": 6079,
                "upload_time": "2018-10-18T02:16:46",
                "url": "https://files.pythonhosted.org/packages/04/e1/6045f77dfb72bdf320b66c0bbf16dc4f35490807c5a63b062de79190329e/repcomp-0.1-py3-none-any.whl"
            },
            {
                "comment_text": "",
                "digests": {
                    "md5": "d2178aa3ad61934d9e9ea01d483dff36",
                    "sha256": "6d742e78699483b67ffdc7a82e000d076bab91ff956b65f64555dd1191c3356f"
                },
                "downloads": -1,
                "filename": "repcomp-0.1.tar.gz",
                "has_sig": false,
                "md5_digest": "d2178aa3ad61934d9e9ea01d483dff36",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 4263,
                "upload_time": "2018-10-18T02:16:46",
                "url": "https://files.pythonhosted.org/packages/e0/a8/8d89424ded043082ee90ef3cc7752d5f144f583f75ae17cd0a873d875a86/repcomp-0.1.tar.gz"
            }
        ]
    },
    "urls": [
        {
            "comment_text": "",
            "digests": {
                "md5": "f301b48f4aa2e11e60bb2c86144c1aa8",
                "sha256": "52184ecd2ffbdbef4587040a71d69ae1301caf8ab450ea357f6f96cfd4415808"
            },
            "downloads": -1,
            "filename": "repcomp-0.1-py3-none-any.whl",
            "has_sig": false,
            "md5_digest": "f301b48f4aa2e11e60bb2c86144c1aa8",
            "packagetype": "bdist_wheel",
            "python_version": "py3",
            "requires_python": null,
            "size": 6079,
            "upload_time": "2018-10-18T02:16:46",
            "url": "https://files.pythonhosted.org/packages/04/e1/6045f77dfb72bdf320b66c0bbf16dc4f35490807c5a63b062de79190329e/repcomp-0.1-py3-none-any.whl"
        },
        {
            "comment_text": "",
            "digests": {
                "md5": "d2178aa3ad61934d9e9ea01d483dff36",
                "sha256": "6d742e78699483b67ffdc7a82e000d076bab91ff956b65f64555dd1191c3356f"
            },
            "downloads": -1,
            "filename": "repcomp-0.1.tar.gz",
            "has_sig": false,
            "md5_digest": "d2178aa3ad61934d9e9ea01d483dff36",
            "packagetype": "sdist",
            "python_version": "source",
            "requires_python": null,
            "size": 4263,
            "upload_time": "2018-10-18T02:16:46",
            "url": "https://files.pythonhosted.org/packages/e0/a8/8d89424ded043082ee90ef3cc7752d5f144f583f75ae17cd0a873d875a86/repcomp-0.1.tar.gz"
        }
    ]
}