{
    "info": {
        "author": "CyberZHG",
        "author_email": "CyberZHG@users.noreply.github.com",
        "bugtrack_url": null,
        "classifiers": [
            "License :: OSI Approved :: MIT License",
            "Operating System :: OS Independent",
            "Programming Language :: Python :: 3"
        ],
        "description": "# Keras Multi-Head\n\n[![Travis](https://travis-ci.org/CyberZHG/keras-multi-head.svg)](https://travis-ci.org/CyberZHG/keras-multi-head)\n[![Coverage](https://coveralls.io/repos/github/CyberZHG/keras-multi-head/badge.svg?branch=master)](https://coveralls.io/github/CyberZHG/keras-multi-head)\n[![Version](https://img.shields.io/pypi/v/keras-multi-head.svg)](https://pypi.org/project/keras-multi-head/)\n![Downloads](https://img.shields.io/pypi/dm/keras-multi-head.svg)\n![License](https://img.shields.io/pypi/l/keras-multi-head.svg)\n\nA wrapper layer for stacking layers horizontally.\n\n![](https://user-images.githubusercontent.com/853842/45797517-867b8580-bcd8-11e8-9ec6-39d6508cf438.png)\n\n## Install\n\n```bash\npip install keras-multi-head\n```\n\n## Usage\n\n### Duplicate Layers\n\nThe layer will be duplicated if only a single layer is provided. The `layer_num` argument controls how many layers will be duplicated eventually.\n\n```python\nimport keras\nfrom keras_multi_head import MultiHead\n\n\nmodel = keras.models.Sequential()\nmodel.add(keras.layers.Embedding(input_dim=100, output_dim=20, name='Embedding'))\nmodel.add(MultiHead(keras.layers.LSTM(units=32), layer_num=5, name='Multi-LSTMs'))\nmodel.add(keras.layers.Flatten(name='Flatten'))\nmodel.add(keras.layers.Dense(units=4, activation='softmax', name='Dense'))\nmodel.build()\nmodel.summary()\n```\n\n### Use Multiple-Layers\n\nThe first argument could also be a list of layers with different configurations, however, they must have the same output shapes.\n\n```python\nimport keras\nfrom keras_multi_head import MultiHead\n\n\nmodel = keras.models.Sequential()\nmodel.add(keras.layers.Embedding(input_dim=100, output_dim=20, name='Embedding'))\nmodel.add(MultiHead([\n    keras.layers.Conv1D(filters=32, kernel_size=3, padding='same'),\n    keras.layers.Conv1D(filters=32, kernel_size=5, padding='same'),\n    keras.layers.Conv1D(filters=32, kernel_size=7, padding='same'),\n], name='Multi-CNNs'))\nmodel.build()\nmodel.summary()\n```\n\n### Linear Transformation\n\nThe input data will be mapped to different values of the same shape for each layer when `hidden_dim` is given.\n\n### Regularization\n\n![](https://user-images.githubusercontent.com/853842/45857922-8b4e4100-bd8d-11e8-905a-4eb07da31418.png)\n\nThe regularization is used when you expect to extract different features from the parallel layers. You can customize the indices of weights in the layers, the intervals represent the parts of the weights and the factor of the regularization.\n\nFor example, the bidirectional LSTM layer has 6 weights by default, and the first 3s belong to the forward layer. The 2nd weight (recurrent kernel) in the forward layer controls the computation of gates for recurrent connections. The kernel for computing cell states lays in units x 2 to units x 3 of the recurrent kernel. We can used the regularization for the kernels:\n\n```python\nimport keras\nfrom keras_multi_head import MultiHead\n\n\nmodel = keras.models.Sequential()\nmodel.add(keras.layers.Embedding(input_dim=5, output_dim=3, name='Embed'))\nmodel.add(MultiHead(\n    layer=keras.layers.Bidirectional(keras.layers.LSTM(units=16), name='LSTM'),\n    layer_num=5,\n    reg_index=[1, 4],\n    reg_slice=(slice(None, None), slice(32, 48)),\n    reg_factor=0.1,\n    name='Multi-Head-Attention',\n))\nmodel.add(keras.layers.Flatten(name='Flatten'))\nmodel.add(keras.layers.Dense(units=2, activation='softmax', name='Dense'))\nmodel.build()\n```\n\n* `reg_index`: The indices of `layer.get_weights()`, a single integer or a list of integers.\n* `reg_slice`: `slice`s or a tuple of `slice`s or a list of the previous choices. If multiple indices are provided in `reg_index` and `reg_slice` is not a list, then `reg_slice` is assumed to be equal for all the indices. The whole array will be used if you leave this argument to `None`.\n* `reg_factor`: The factor of the regularization, a float or a list of floats.\n\n### Multi-Head Attention\n\nA more specific multi-head layer is provided (since the general one is harder to use). The layer uses scaled dot product attention layers as its sub-layers and only `head_num` is required:\n\n```python\nimport keras\nfrom keras_multi_head import MultiHeadAttention\n\ninput_layer = keras.layers.Input(\n    shape=(2, 3),\n    name='Input',\n)\natt_layer = MultiHeadAttention(\n    head_num=3,\n    name='Multi-Head',\n)(input_layer)\nmodel = keras.models.Model(inputs=input_layer, outputs=att_layer)\nmodel.compile(\n    optimizer='adam',\n    loss='mse',\n    metrics={},\n)\nmodel.summary()\n```\n\nThe shapes of input and output tensors would be the same if only one layer is presented as input. The input layers will be considered as query, key and value when a list is given:\n\n```python\nimport keras\nfrom keras_multi_head import MultiHeadAttention\n\ninput_query = keras.layers.Input(\n    shape=(2, 3),\n    name='Input-Q',\n)\ninput_key = keras.layers.Input(\n    shape=(4, 5),\n    name='Input-K',\n)\ninput_value = keras.layers.Input(\n    shape=(4, 6),\n    name='Input-V',\n)\natt_layer = MultiHeadAttention(\n    head_num=3,\n    name='Multi-Head',\n)([input_query, input_key, input_value])\nmodel = keras.models.Model(inputs=[input_query, input_key, input_value], outputs=att_layer)\nmodel.compile(\n    optimizer='adam',\n    loss='mse',\n    metrics={},\n)\nmodel.summary()\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/CyberZHG/keras-multi-head",
        "keywords": "",
        "license": "MIT",
        "maintainer": "",
        "maintainer_email": "",
        "name": "keras-multi-head",
        "package_url": "https://pypi.org/project/keras-multi-head/",
        "platform": "",
        "project_url": "https://pypi.org/project/keras-multi-head/",
        "project_urls": {
            "Homepage": "https://github.com/CyberZHG/keras-multi-head"
        },
        "release_url": "https://pypi.org/project/keras-multi-head/0.22.0/",
        "requires_dist": null,
        "requires_python": "",
        "summary": "A wrapper layer for stacking layers horizontally",
        "version": "0.22.0"
    },
    "last_serial": 5718862,
    "releases": {
        "0.1": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "520349d7da205b589324d1de4dd4abab",
                    "sha256": "2a940a2fc85262ceb6292dcab998d396cca11f938cee12f62a4d7064cc539b6f"
                },
                "downloads": -1,
                "filename": "keras-multi-head-0.1.tar.gz",
                "has_sig": false,
                "md5_digest": "520349d7da205b589324d1de4dd4abab",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 1941,
                "upload_time": "2018-09-20T02:59:09",
                "url": "https://files.pythonhosted.org/packages/b6/89/138e4c1c440f470f5c4eb4f18733a671097f37534b90ad754a83e7987ccd/keras-multi-head-0.1.tar.gz"
            }
        ],
        "0.10.0": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "38d78b08f7e6e7674500329af36510d3",
                    "sha256": "c242189663626020465d7436e7792b2dd963d45a64ed35b56257670fafa8a0db"
                },
                "downloads": -1,
                "filename": "keras-multi-head-0.10.0.tar.gz",
                "has_sig": false,
                "md5_digest": "38d78b08f7e6e7674500329af36510d3",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 6688,
                "upload_time": "2018-11-09T05:16:20",
                "url": "https://files.pythonhosted.org/packages/65/fb/ae3f3835915c725010233eb175a47af0b4f06f12b7f5cff1a3daf362cff1/keras-multi-head-0.10.0.tar.gz"
            }
        ],
        "0.11.0": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "dd29971421ff29419e73c709687399b4",
                    "sha256": "c88bc9c5f5e4406e7de80fc59c960a109565c4ee1547f8ce560150f00017a9b1"
                },
                "downloads": -1,
                "filename": "keras-multi-head-0.11.0.tar.gz",
                "has_sig": false,
                "md5_digest": "dd29971421ff29419e73c709687399b4",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 6739,
                "upload_time": "2018-11-12T06:12:32",
                "url": "https://files.pythonhosted.org/packages/05/a8/9800d6fa5fc9bbd56cde02831d0b393ca3d9bf42e3d375ace5b819d694fb/keras-multi-head-0.11.0.tar.gz"
            }
        ],
        "0.12.0": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "7823a29507e2d4989e5edfc403910fd5",
                    "sha256": "632801ad744438fd5d86bca52074170ff04ea2153cab4b05ac812c432724a1ea"
                },
                "downloads": -1,
                "filename": "keras-multi-head-0.12.0.tar.gz",
                "has_sig": false,
                "md5_digest": "7823a29507e2d4989e5edfc403910fd5",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 6742,
                "upload_time": "2018-11-12T06:45:10",
                "url": "https://files.pythonhosted.org/packages/64/af/ce3744a0813334b1e351f86dee0a8909b4ef7042d154afba878b6e31588d/keras-multi-head-0.12.0.tar.gz"
            }
        ],
        "0.13.0": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "60541b313842e7ee43f16461b3dc33b5",
                    "sha256": "c0d613e2f0ca1a810d29ce3264d546ed8fc05d8c5abe53e5c0f0bd30abd568ac"
                },
                "downloads": -1,
                "filename": "keras-multi-head-0.13.0.tar.gz",
                "has_sig": false,
                "md5_digest": "60541b313842e7ee43f16461b3dc33b5",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 7006,
                "upload_time": "2018-11-12T09:43:29",
                "url": "https://files.pythonhosted.org/packages/08/1d/efd262bb9ee8b3c67b34e63eb2ac7821f88d866e1c72638136789e096a14/keras-multi-head-0.13.0.tar.gz"
            }
        ],
        "0.14.0": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "d1d3edecd0d1e4b68dcc7e5cde19465f",
                    "sha256": "d3dcfbbe06b29846d04c2912ebdccf5080d5de8da9b0da7543a426415a443161"
                },
                "downloads": -1,
                "filename": "keras-multi-head-0.14.0.tar.gz",
                "has_sig": false,
                "md5_digest": "d1d3edecd0d1e4b68dcc7e5cde19465f",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 7194,
                "upload_time": "2018-11-13T02:50:59",
                "url": "https://files.pythonhosted.org/packages/86/95/3c5dea5a63beb60985a9a29fb883c8849ebfb5a8f20a790bfd22df24543c/keras-multi-head-0.14.0.tar.gz"
            }
        ],
        "0.15.0": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "00ae1bb3f1d9caf0a74e0c104ac8bd88",
                    "sha256": "ea9ebe2297666ea8cf004369b179fe607a331026ba76158d99761a3e5a1417ed"
                },
                "downloads": -1,
                "filename": "keras-multi-head-0.15.0.tar.gz",
                "has_sig": false,
                "md5_digest": "00ae1bb3f1d9caf0a74e0c104ac8bd88",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 11971,
                "upload_time": "2019-02-01T03:10:16",
                "url": "https://files.pythonhosted.org/packages/c0/31/240d4e125e4a96a40c490e861679c2821992f0836fc193c1765b353eaf07/keras-multi-head-0.15.0.tar.gz"
            }
        ],
        "0.16.0": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "87a7be09644d9a92586f8cb04f2b2203",
                    "sha256": "e5ee5c10c32eeaa9a3688103cc91f3ef36973eff66d14b21c6ffcf72227a3b04"
                },
                "downloads": -1,
                "filename": "keras-multi-head-0.16.0.tar.gz",
                "has_sig": false,
                "md5_digest": "87a7be09644d9a92586f8cb04f2b2203",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 11965,
                "upload_time": "2019-03-11T06:17:25",
                "url": "https://files.pythonhosted.org/packages/7e/c9/7811e06a523c743d1a7449e8647d2ef1c7713248e27ea6f6b23772d38881/keras-multi-head-0.16.0.tar.gz"
            }
        ],
        "0.17.0": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "2c472820d555012cb23ec39e96c93db9",
                    "sha256": "da5f8383ffb6a4528bbb9f5c5b32bac704e0cb24e6178b95b35eaf3d6543f139"
                },
                "downloads": -1,
                "filename": "keras-multi-head-0.17.0.tar.gz",
                "has_sig": false,
                "md5_digest": "2c472820d555012cb23ec39e96c93db9",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 11946,
                "upload_time": "2019-04-16T07:31:29",
                "url": "https://files.pythonhosted.org/packages/c0/1d/9299152cf06956dd825404ba6361500f2b0642e98336f802fd89d8dab80f/keras-multi-head-0.17.0.tar.gz"
            }
        ],
        "0.18.0": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "38dde214879b7f512ff716946f1d924c",
                    "sha256": "6adc3c743f6c45e287f2a96d491220fc13da9a8c0a497d9428a54a531c70e4d7"
                },
                "downloads": -1,
                "filename": "keras-multi-head-0.18.0.tar.gz",
                "has_sig": false,
                "md5_digest": "38dde214879b7f512ff716946f1d924c",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 11955,
                "upload_time": "2019-04-16T08:34:03",
                "url": "https://files.pythonhosted.org/packages/d7/5d/8156def9ca75c55bb87819618e9a3e1f8e587c722570e2e93ad616b9269d/keras-multi-head-0.18.0.tar.gz"
            }
        ],
        "0.19.0": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "02cc93e51b0ad34e8711af6404347676",
                    "sha256": "fc0e5906b198c402217e1c1926bff9355e42235d8fd5e48c857d28a253e11e2b"
                },
                "downloads": -1,
                "filename": "keras-multi-head-0.19.0.tar.gz",
                "has_sig": false,
                "md5_digest": "02cc93e51b0ad34e8711af6404347676",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 10685,
                "upload_time": "2019-05-11T14:15:12",
                "url": "https://files.pythonhosted.org/packages/80/63/f242f120b5b0a35d5e2601f35c7578440fe06f5d3821308093080cb50907/keras-multi-head-0.19.0.tar.gz"
            }
        ],
        "0.2": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "18bdc65019101c6a6154266f3d6cc7f6",
                    "sha256": "89b45db4f69febf1e44421831083af142380b469b24badb41081cdc5412196b2"
                },
                "downloads": -1,
                "filename": "keras-multi-head-0.2.tar.gz",
                "has_sig": false,
                "md5_digest": "18bdc65019101c6a6154266f3d6cc7f6",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 2558,
                "upload_time": "2018-09-20T04:41:37",
                "url": "https://files.pythonhosted.org/packages/74/7a/856577d2dc34cd5ee5d6b2839f6f409ab4940aae9616005bed2bc3a57fa4/keras-multi-head-0.2.tar.gz"
            }
        ],
        "0.20.0": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "3e104089199ec25832b164659281ee70",
                    "sha256": "5ce199e9a075b14956e714420a7f36baaa66397a9c85e62d6c342cf9afe9c265"
                },
                "downloads": -1,
                "filename": "keras-multi-head-0.20.0.tar.gz",
                "has_sig": false,
                "md5_digest": "3e104089199ec25832b164659281ee70",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 13790,
                "upload_time": "2019-05-21T10:04:24",
                "url": "https://files.pythonhosted.org/packages/3b/0f/f1a66974db9c328ba675c1df63f8a68c5c0f3e181f1e74db4f3b1a72a6df/keras-multi-head-0.20.0.tar.gz"
            }
        ],
        "0.22.0": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "8739cd467ad673ac60e6eae6aed36c7c",
                    "sha256": "bce7e10ee2b5f58a6c0c493dfae3b08ed6c1771ac48e68cb466d6871b610427d"
                },
                "downloads": -1,
                "filename": "keras-multi-head-0.22.0.tar.gz",
                "has_sig": false,
                "md5_digest": "8739cd467ad673ac60e6eae6aed36c7c",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 12139,
                "upload_time": "2019-08-23T05:01:22",
                "url": "https://files.pythonhosted.org/packages/40/3e/d0a64bb2ac5217928effe4507c26bbd19b86145d16a1948bc2d4f4c6338a/keras-multi-head-0.22.0.tar.gz"
            }
        ],
        "0.3": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "b6b681aa4d634bf0812ec5542e06e573",
                    "sha256": "5f9eb9dfcf04f13b394e93f86750d80befe0ae1180b6043ab08315c052ec3457"
                },
                "downloads": -1,
                "filename": "keras-multi-head-0.3.tar.gz",
                "has_sig": false,
                "md5_digest": "b6b681aa4d634bf0812ec5542e06e573",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 3182,
                "upload_time": "2018-09-20T05:50:35",
                "url": "https://files.pythonhosted.org/packages/20/97/0c4c1b161cfdd73618e19b84b82cfa02b3b88e3e7980c1b034b7a8598dcd/keras-multi-head-0.3.tar.gz"
            }
        ],
        "0.4": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "bbfc04c8b6eab135cbe6f20f5b30c2f3",
                    "sha256": "ce72e4197ed8292d76d6bd230259336bb2b35898604ff748186ad167a41cd622"
                },
                "downloads": -1,
                "filename": "keras-multi-head-0.4.tar.gz",
                "has_sig": false,
                "md5_digest": "bbfc04c8b6eab135cbe6f20f5b30c2f3",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 4467,
                "upload_time": "2018-09-21T04:09:26",
                "url": "https://files.pythonhosted.org/packages/79/a3/d354bec0da95d4ef79c0a1de8372bc9508df948cda8063830dba4e4d3645/keras-multi-head-0.4.tar.gz"
            }
        ],
        "0.5": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "2b75ea19834b8803e64cbb4c36031015",
                    "sha256": "197202e9acef8b720655bb9ed3616280933454790df03390c04459bc052b0cf3"
                },
                "downloads": -1,
                "filename": "keras-multi-head-0.5.tar.gz",
                "has_sig": false,
                "md5_digest": "2b75ea19834b8803e64cbb4c36031015",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 4849,
                "upload_time": "2018-10-19T10:56:10",
                "url": "https://files.pythonhosted.org/packages/8b/fc/b404907b804d38ebb02ce14ceae6636e1ebbe372dbd13e6070b3a63a1d2e/keras-multi-head-0.5.tar.gz"
            }
        ],
        "0.6.0": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "caad572ea9ce146ba249f3be763dedbc",
                    "sha256": "e8c9c133ac82c59917cf9610d609f4549ff14b32f835c78fa18ed545fad05a7b"
                },
                "downloads": -1,
                "filename": "keras-multi-head-0.6.0.tar.gz",
                "has_sig": false,
                "md5_digest": "caad572ea9ce146ba249f3be763dedbc",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 5914,
                "upload_time": "2018-11-07T08:20:39",
                "url": "https://files.pythonhosted.org/packages/22/a2/aa4f7574a3c585ad836b145a66c14ae13959d543809795482452a304e0d0/keras-multi-head-0.6.0.tar.gz"
            }
        ],
        "0.7.0": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "2d9d74a0ad561daad173c2c6e3b04681",
                    "sha256": "13586ac8b4ebe760df25d9f71edafcc00b8537b1112b8e8c5e86fc14d32da1f1"
                },
                "downloads": -1,
                "filename": "keras-multi-head-0.7.0.tar.gz",
                "has_sig": false,
                "md5_digest": "2d9d74a0ad561daad173c2c6e3b04681",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 5925,
                "upload_time": "2018-11-07T09:07:24",
                "url": "https://files.pythonhosted.org/packages/2c/7f/ea395f42a284363ad027fa2ad07727c216eb62d87db4b88f4c948d6a80c5/keras-multi-head-0.7.0.tar.gz"
            }
        ],
        "0.8.0": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "2ecaed81a2140f8dfc794f7b8b0e6d9b",
                    "sha256": "86437cde6ea31c89d7bf52b84d8ff75e52b39f0216df20794e6bddbc957e9400"
                },
                "downloads": -1,
                "filename": "keras-multi-head-0.8.0.tar.gz",
                "has_sig": false,
                "md5_digest": "2ecaed81a2140f8dfc794f7b8b0e6d9b",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 5965,
                "upload_time": "2018-11-08T08:28:18",
                "url": "https://files.pythonhosted.org/packages/96/29/7104b2b1d03a4e1e92d4cb9086f45b8acd0b05206a5188f652689b17897b/keras-multi-head-0.8.0.tar.gz"
            }
        ],
        "0.9.0": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "91058449f517b841da88e8a41c1a5592",
                    "sha256": "71068b5e369eb105427c9ebc1f4740926ad2d29df66260d253b952eca1bc2c4d"
                },
                "downloads": -1,
                "filename": "keras-multi-head-0.9.0.tar.gz",
                "has_sig": false,
                "md5_digest": "91058449f517b841da88e8a41c1a5592",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 6645,
                "upload_time": "2018-11-08T11:13:00",
                "url": "https://files.pythonhosted.org/packages/e3/7c/6cbb64f969a44ade973adbed5b7419e49f35a04ab66a4f049a4c4484d6ec/keras-multi-head-0.9.0.tar.gz"
            }
        ]
    },
    "urls": [
        {
            "comment_text": "",
            "digests": {
                "md5": "8739cd467ad673ac60e6eae6aed36c7c",
                "sha256": "bce7e10ee2b5f58a6c0c493dfae3b08ed6c1771ac48e68cb466d6871b610427d"
            },
            "downloads": -1,
            "filename": "keras-multi-head-0.22.0.tar.gz",
            "has_sig": false,
            "md5_digest": "8739cd467ad673ac60e6eae6aed36c7c",
            "packagetype": "sdist",
            "python_version": "source",
            "requires_python": null,
            "size": 12139,
            "upload_time": "2019-08-23T05:01:22",
            "url": "https://files.pythonhosted.org/packages/40/3e/d0a64bb2ac5217928effe4507c26bbd19b86145d16a1948bc2d4f4c6338a/keras-multi-head-0.22.0.tar.gz"
        }
    ]
}