{
    "info": {
        "author": "Example Author",
        "author_email": "author@example.com",
        "bugtrack_url": null,
        "classifiers": [
            "License :: OSI Approved :: MIT License",
            "Operating System :: OS Independent",
            "Programming Language :: Python :: 3"
        ],
        "description": "# Text Classification with CNN and RNN\n\n### \u9700\u8981\u4fee\u6539\u5730\u65b9\n1\uff0c\u9700\u8981\u4fee\u6539\u5355\u8bcd\u5b57\u5178\uff1b  \n2\uff0c\u6587\u4ef6\u8bfb\u53d6\u4e0d\u6210\u529f\n3\uff0c\u4fee\u6539cnn\u6a21\u578b\u4e2d\u7684\u7c7b\u522b\u6570\u76ee\n----\nupdate\u8fd9\u4e48\u591a\u4e1c\u897f\u53bb\u54ea\u91cc\u4e86 CNN\u505a\u53e5\u5b50\u5206\u7c7b\u7684\u8bba\u6587\u53ef\u4ee5\u53c2\u770b: [Convolutional Neural Networks for Sentence Classification](https://arxiv.org/abs/1408.5882)\n\n\u8fd8\u53ef\u4ee5\u53bb\u8bfbdennybritz\u5927\u725b\u7684\u535a\u5ba2\uff1a[Implementing a CNN for Text Classification in TensorFlow](http://www.wildml.com/2015/12/implementing-a-cnn-for-text-classification-in-tensorflow/)\n\n\u4ee5\u53ca\u5b57\u7b26\u7ea7CNN\u7684\u8bba\u6587\uff1a[Character-level Convolutional Networks for Text Classification](https://arxiv.org/abs/1509.01626)\n\n\u672c\u6587\u662f\u57fa\u4e8eTensorFlow\u5728\u4e2d\u6587\u6570\u636e\u96c6\u4e0a\u7684\u7b80\u5316\u5b9e\u73b0\uff0c\u4f7f\u7528\u4e86\u5b57\u7b26\u7ea7CNN\u548cRNN\u5bf9\u4e2d\u6587\u6587\u672c\u8fdb\u884c\u5206\u7c7b\uff0c\u8fbe\u5230\u4e86\u8f83\u597d\u7684\u6548\u679c\u3002\n\n\u6587\u4e2d\u6240\u4f7f\u7528\u7684Conv1D\u4e0e\u8bba\u6587\u4e2d\u6709\u4e9b\u4e0d\u540c\uff0c\u8be6\u7ec6\u53c2\u8003\u5b98\u65b9\u6587\u6863\uff1a[tf.nn.conv1d](https://www.tensorflow.org/api_docs/python/tf/nn/conv1d)\n\n## \u73af\u5883\n\n- Python 2/3 (\u611f\u8c22[howie.hu](https://github.com/howie6879)\u8c03\u8bd5Python2\u73af\u5883)\n- TensorFlow 1.3\u4ee5\u4e0a\n- numpy\n- scikit-learn\n- scipy\n\n## \u6570\u636e\u96c6\n\n\u4f7f\u7528THUCNews\u7684\u4e00\u4e2a\u5b50\u96c6\u8fdb\u884c\u8bad\u7ec3\u4e0e\u6d4b\u8bd5\uff0c\u6570\u636e\u96c6\u8bf7\u81ea\u884c\u5230[THUCTC\uff1a\u4e00\u4e2a\u9ad8\u6548\u7684\u4e2d\u6587\u6587\u672c\u5206\u7c7b\u5de5\u5177\u5305](http://thuctc.thunlp.org/)\u4e0b\u8f7d\uff0c\u8bf7\u9075\u5faa\u6570\u636e\u63d0\u4f9b\u65b9\u7684\u5f00\u6e90\u534f\u8bae\u3002\n\n\u672c\u6b21\u8bad\u7ec3\u4f7f\u7528\u4e86\u5176\u4e2d\u768410\u4e2a\u5206\u7c7b\uff0c\u6bcf\u4e2a\u5206\u7c7b6500\u6761\u6570\u636e\u3002\n\n\u7c7b\u522b\u5982\u4e0b\uff1a\n\n```\n\u4f53\u80b2, \u8d22\u7ecf, \u623f\u4ea7, \u5bb6\u5c45, \u6559\u80b2, \u79d1\u6280, \u65f6\u5c1a, \u65f6\u653f, \u6e38\u620f, \u5a31\u4e50\n```\n\n\u8fd9\u4e2a\u5b50\u96c6\u53ef\u4ee5\u5728\u6b64\u4e0b\u8f7d\uff1a\u94fe\u63a5: https://pan.baidu.com/s/1hugrfRu \u5bc6\u7801: qfud\n\n\u6570\u636e\u96c6\u5212\u5206\u5982\u4e0b\uff1a\n\n- \u8bad\u7ec3\u96c6: 5000*10\n- \u9a8c\u8bc1\u96c6: 500*10\n- \u6d4b\u8bd5\u96c6: 1000*10\n\n\u4ece\u539f\u6570\u636e\u96c6\u751f\u6210\u5b50\u96c6\u7684\u8fc7\u7a0b\u8bf7\u53c2\u770b`helper`\u4e0b\u7684\u4e24\u4e2a\u811a\u672c\u3002\u5176\u4e2d\uff0c`copy_data.sh`\u7528\u4e8e\u4ece\u6bcf\u4e2a\u5206\u7c7b\u62f7\u8d1d6500\u4e2a\u6587\u4ef6\uff0c`cnews_group.py`\u7528\u4e8e\u5c06\u591a\u4e2a\u6587\u4ef6\u6574\u5408\u5230\u4e00\u4e2a\u6587\u4ef6\u4e2d\u3002\u6267\u884c\u8be5\u6587\u4ef6\u540e\uff0c\u5f97\u5230\u4e09\u4e2a\u6570\u636e\u6587\u4ef6\uff1a\n\n- cnews.train.txt: \u8bad\u7ec3\u96c6(50000\u6761)\n- cnews.val.txt: \u9a8c\u8bc1\u96c6(5000\u6761)\n- cnews.test.txt: \u6d4b\u8bd5\u96c6(10000\u6761)\n\n## \u9884\u5904\u7406\n\n`data/cnews_loader.py`\u4e3a\u6570\u636e\u7684\u9884\u5904\u7406\u6587\u4ef6\u3002\n\n- `read_file()`: \u8bfb\u53d6\u6587\u4ef6\u6570\u636e;\n- `build_vocab()`: \u6784\u5efa\u8bcd\u6c47\u8868\uff0c\u4f7f\u7528\u5b57\u7b26\u7ea7\u7684\u8868\u793a\uff0c\u8fd9\u4e00\u51fd\u6570\u4f1a\u5c06\u8bcd\u6c47\u8868\u5b58\u50a8\u4e0b\u6765\uff0c\u907f\u514d\u6bcf\u4e00\u6b21\u91cd\u590d\u5904\u7406;\n- `read_vocab()`: \u8bfb\u53d6\u4e0a\u4e00\u6b65\u5b58\u50a8\u7684\u8bcd\u6c47\u8868\uff0c\u8f6c\u6362\u4e3a`{\u8bcd\uff1aid}`\u8868\u793a;\n- `read_category()`: \u5c06\u5206\u7c7b\u76ee\u5f55\u56fa\u5b9a\uff0c\u8f6c\u6362\u4e3a`{\u7c7b\u522b: id}`\u8868\u793a;\n- `to_words()`: \u5c06\u4e00\u6761\u7531id\u8868\u793a\u7684\u6570\u636e\u91cd\u65b0\u8f6c\u6362\u4e3a\u6587\u5b57;\n- `process_file()`: \u5c06\u6570\u636e\u96c6\u4ece\u6587\u5b57\u8f6c\u6362\u4e3a\u56fa\u5b9a\u957f\u5ea6\u7684id\u5e8f\u5217\u8868\u793a;\n- `batch_iter()`: \u4e3a\u795e\u7ecf\u7f51\u7edc\u7684\u8bad\u7ec3\u51c6\u5907\u7ecf\u8fc7shuffle\u7684\u6279\u6b21\u7684\u6570\u636e\u3002\n\n\u7ecf\u8fc7\u6570\u636e\u9884\u5904\u7406\uff0c\u6570\u636e\u7684\u683c\u5f0f\u5982\u4e0b\uff1a\n\n| Data | Shape | Data | Shape |\n| :---------- | :---------- | :---------- | :---------- |\n| x_train | [50000, 600] | y_train | [50000, 10] |\n| x_val | [5000, 600] | y_val | [5000, 10] |\n| x_test | [10000, 600] | y_test | [10000, 10] |\n\n## CNN\u5377\u79ef\u795e\u7ecf\u7f51\u7edc\n\n### \u914d\u7f6e\u9879\n\nCNN\u53ef\u914d\u7f6e\u7684\u53c2\u6570\u5982\u4e0b\u6240\u793a\uff0c\u5728`cnn_model.py`\u4e2d\u3002\n\n```python\nclass TCNNConfig(object):\n    \"\"\"CNN\u914d\u7f6e\u53c2\u6570\"\"\"\n\n    embedding_dim = 64      # \u8bcd\u5411\u91cf\u7ef4\u5ea6\n    seq_length = 600        # \u5e8f\u5217\u957f\u5ea6\n    num_classes = 10        # \u7c7b\u522b\u6570\n    num_filters = 128        # \u5377\u79ef\u6838\u6570\u76ee\n    kernel_size = 5         # \u5377\u79ef\u6838\u5c3a\u5bf8\n    vocab_size = 5000       # \u8bcd\u6c47\u8868\u8fbe\u5c0f\n\n    hidden_dim = 128        # \u5168\u8fde\u63a5\u5c42\u795e\u7ecf\u5143\n\n    dropout_keep_prob = 0.5 # dropout\u4fdd\u7559\u6bd4\u4f8b\n    learning_rate = 1e-3    # \u5b66\u4e60\u7387\n\n    batch_size = 64         # \u6bcf\u6279\u8bad\u7ec3\u5927\u5c0f\n    num_epochs = 10         # \u603b\u8fed\u4ee3\u8f6e\u6b21\n\n    print_per_batch = 100    # \u6bcf\u591a\u5c11\u8f6e\u8f93\u51fa\u4e00\u6b21\u7ed3\u679c\n    save_per_batch = 10      # \u6bcf\u591a\u5c11\u8f6e\u5b58\u5165tensorboard\n```\n\n### CNN\u6a21\u578b\n\n\u5177\u4f53\u53c2\u770b`cnn_model.py`\u7684\u5b9e\u73b0\u3002\n\n\u5927\u81f4\u7ed3\u6784\u5982\u4e0b\uff1a\n\n![images/cnn_architecture](images/cnn_architecture.png)\n\n### \u8bad\u7ec3\u4e0e\u9a8c\u8bc1\n\n\u8fd0\u884c `python run_cnn.py train`\uff0c\u53ef\u4ee5\u5f00\u59cb\u8bad\u7ec3\u3002\n\n> \u82e5\u4e4b\u524d\u8fdb\u884c\u8fc7\u8bad\u7ec3\uff0c\u8bf7\u628atensorboard/textcnn\u5220\u9664\uff0c\u907f\u514dTensorBoard\u591a\u6b21\u8bad\u7ec3\u7ed3\u679c\u91cd\u53e0\u3002\n\n```\nConfiguring CNN model...\nConfiguring TensorBoard and Saver...\nLoading training and validation data...\nTime usage: 0:00:14\nTraining and evaluating...\nEpoch: 1\nIter:      0, Train Loss:    2.3, Train Acc:  10.94%, Val Loss:    2.3, Val Acc:   8.92%, Time: 0:00:01 *\nIter:    100, Train Loss:   0.88, Train Acc:  73.44%, Val Loss:    1.2, Val Acc:  68.46%, Time: 0:00:04 *\nIter:    200, Train Loss:   0.38, Train Acc:  92.19%, Val Loss:   0.75, Val Acc:  77.32%, Time: 0:00:07 *\nIter:    300, Train Loss:   0.22, Train Acc:  92.19%, Val Loss:   0.46, Val Acc:  87.08%, Time: 0:00:09 *\nIter:    400, Train Loss:   0.24, Train Acc:  90.62%, Val Loss:    0.4, Val Acc:  88.62%, Time: 0:00:12 *\nIter:    500, Train Loss:   0.16, Train Acc:  96.88%, Val Loss:   0.36, Val Acc:  90.38%, Time: 0:00:15 *\nIter:    600, Train Loss:  0.084, Train Acc:  96.88%, Val Loss:   0.35, Val Acc:  91.36%, Time: 0:00:17 *\nIter:    700, Train Loss:   0.21, Train Acc:  93.75%, Val Loss:   0.26, Val Acc:  92.58%, Time: 0:00:20 *\nEpoch: 2\nIter:    800, Train Loss:   0.07, Train Acc:  98.44%, Val Loss:   0.24, Val Acc:  94.12%, Time: 0:00:23 *\nIter:    900, Train Loss:  0.092, Train Acc:  96.88%, Val Loss:   0.27, Val Acc:  92.86%, Time: 0:00:25\nIter:   1000, Train Loss:   0.17, Train Acc:  95.31%, Val Loss:   0.28, Val Acc:  92.82%, Time: 0:00:28\nIter:   1100, Train Loss:    0.2, Train Acc:  93.75%, Val Loss:   0.23, Val Acc:  93.26%, Time: 0:00:31\nIter:   1200, Train Loss:  0.081, Train Acc:  98.44%, Val Loss:   0.25, Val Acc:  92.96%, Time: 0:00:33\nIter:   1300, Train Loss:  0.052, Train Acc: 100.00%, Val Loss:   0.24, Val Acc:  93.58%, Time: 0:00:36\nIter:   1400, Train Loss:    0.1, Train Acc:  95.31%, Val Loss:   0.22, Val Acc:  94.12%, Time: 0:00:39\nIter:   1500, Train Loss:   0.12, Train Acc:  98.44%, Val Loss:   0.23, Val Acc:  93.58%, Time: 0:00:41\nEpoch: 3\nIter:   1600, Train Loss:    0.1, Train Acc:  96.88%, Val Loss:   0.26, Val Acc:  92.34%, Time: 0:00:44\nIter:   1700, Train Loss:  0.018, Train Acc: 100.00%, Val Loss:   0.22, Val Acc:  93.46%, Time: 0:00:47\nIter:   1800, Train Loss:  0.036, Train Acc: 100.00%, Val Loss:   0.28, Val Acc:  92.72%, Time: 0:00:50\nNo optimization for a long time, auto-stopping...\n```\n\n\u5728\u9a8c\u8bc1\u96c6\u4e0a\u7684\u6700\u4f73\u6548\u679c\u4e3a94.12%\uff0c\u4e14\u53ea\u7ecf\u8fc7\u4e863\u8f6e\u8fed\u4ee3\u5c31\u5df2\u7ecf\u505c\u6b62\u3002\n\n\u51c6\u786e\u7387\u548c\u8bef\u5dee\u5982\u56fe\u6240\u793a\uff1a\n\n![images](images/acc_loss.png)\n\n\n### \u6d4b\u8bd5\n\n\u8fd0\u884c `python run_cnn.py test` \u5728\u6d4b\u8bd5\u96c6\u4e0a\u8fdb\u884c\u6d4b\u8bd5\u3002\n\n```\nConfiguring CNN model...\nLoading test data...\nTesting...\nTest Loss:   0.14, Test Acc:  96.04%\nPrecision, Recall and F1-Score...\n             precision    recall  f1-score   support\n\n         \u4f53\u80b2       0.99      0.99      0.99      1000\n         \u8d22\u7ecf       0.96      0.99      0.97      1000\n         \u623f\u4ea7       1.00      1.00      1.00      1000\n         \u5bb6\u5c45       0.95      0.91      0.93      1000\n         \u6559\u80b2       0.95      0.89      0.92      1000\n         \u79d1\u6280       0.94      0.97      0.95      1000\n         \u65f6\u5c1a       0.95      0.97      0.96      1000\n         \u65f6\u653f       0.94      0.94      0.94      1000\n         \u6e38\u620f       0.97      0.96      0.97      1000\n         \u5a31\u4e50       0.95      0.98      0.97      1000\n\navg / total       0.96      0.96      0.96     10000\n\nConfusion Matrix...\n[[991   0   0   0   2   1   0   4   1   1]\n [  0 992   0   0   2   1   0   5   0   0]\n [  0   1 996   0   1   1   0   0   0   1]\n [  0  14   0 912   7  15   9  29   3  11]\n [  2   9   0  12 892  22  18  21  10  14]\n [  0   0   0  10   1 968   4   3  12   2]\n [  1   0   0   9   4   4 971   0   2   9]\n [  1  16   0   4  18  12   1 941   1   6]\n [  2   4   1   5   4   5  10   1 962   6]\n [  1   0   1   6   4   3   5   0   1 979]]\nTime usage: 0:00:05\n```\n\n\u5728\u6d4b\u8bd5\u96c6\u4e0a\u7684\u51c6\u786e\u7387\u8fbe\u5230\u4e8696.04%\uff0c\u4e14\u5404\u7c7b\u7684precision, recall\u548cf1-score\u90fd\u8d85\u8fc7\u4e860.9\u3002\n\n\u4ece\u6df7\u6dc6\u77e9\u9635\u4e5f\u53ef\u4ee5\u770b\u51fa\u5206\u7c7b\u6548\u679c\u975e\u5e38\u4f18\u79c0\u3002\n\n## RNN\u5faa\u73af\u795e\u7ecf\u7f51\u7edc\n\n### \u914d\u7f6e\u9879\n\nRNN\u53ef\u914d\u7f6e\u7684\u53c2\u6570\u5982\u4e0b\u6240\u793a\uff0c\u5728`rnn_model.py`\u4e2d\u3002\n\n```python\nclass TRNNConfig(object):\n    \"\"\"RNN\u914d\u7f6e\u53c2\u6570\"\"\"\n\n    # \u6a21\u578b\u53c2\u6570\n    embedding_dim = 64      # \u8bcd\u5411\u91cf\u7ef4\u5ea6\n    seq_length = 600        # \u5e8f\u5217\u957f\u5ea6\n    num_classes = 10        # \u7c7b\u522b\u6570\n    vocab_size = 5000       # \u8bcd\u6c47\u8868\u8fbe\u5c0f\n\n    num_layers= 2           # \u9690\u85cf\u5c42\u5c42\u6570\n    hidden_dim = 128        # \u9690\u85cf\u5c42\u795e\u7ecf\u5143\n    rnn = 'gru'             # lstm \u6216 gru\n\n    dropout_keep_prob = 0.8 # dropout\u4fdd\u7559\u6bd4\u4f8b\n    learning_rate = 1e-3    # \u5b66\u4e60\u7387\n\n    batch_size = 128         # \u6bcf\u6279\u8bad\u7ec3\u5927\u5c0f\n    num_epochs = 10          # \u603b\u8fed\u4ee3\u8f6e\u6b21\n\n    print_per_batch = 100    # \u6bcf\u591a\u5c11\u8f6e\u8f93\u51fa\u4e00\u6b21\u7ed3\u679c\n    save_per_batch = 10      # \u6bcf\u591a\u5c11\u8f6e\u5b58\u5165tensorboard\n```\n\n### RNN\u6a21\u578b\n\n\u5177\u4f53\u53c2\u770b`rnn_model.py`\u7684\u5b9e\u73b0\u3002\n\n\u5927\u81f4\u7ed3\u6784\u5982\u4e0b\uff1a\n\n![images/rnn_architecture](images/rnn_architecture.png)\n\n### \u8bad\u7ec3\u4e0e\u9a8c\u8bc1\n\n> \u8fd9\u90e8\u5206\u7684\u4ee3\u7801\u4e0e run_cnn.py\u6781\u4e3a\u76f8\u4f3c\uff0c\u53ea\u9700\u8981\u5c06\u6a21\u578b\u548c\u90e8\u5206\u76ee\u5f55\u7a0d\u5fae\u4fee\u6539\u3002\n\n\u8fd0\u884c `python run_rnn.py train`\uff0c\u53ef\u4ee5\u5f00\u59cb\u8bad\u7ec3\u3002\n\n> \u82e5\u4e4b\u524d\u8fdb\u884c\u8fc7\u8bad\u7ec3\uff0c\u8bf7\u628atensorboard/textrnn\u5220\u9664\uff0c\u907f\u514dTensorBoard\u591a\u6b21\u8bad\u7ec3\u7ed3\u679c\u91cd\u53e0\u3002\n\n```\nConfiguring RNN model...\nConfiguring TensorBoard and Saver...\nLoading training and validation data...\nTime usage: 0:00:14\nTraining and evaluating...\nEpoch: 1\nIter:      0, Train Loss:    2.3, Train Acc:   8.59%, Val Loss:    2.3, Val Acc:  11.96%, Time: 0:00:08 *\nIter:    100, Train Loss:   0.95, Train Acc:  64.06%, Val Loss:    1.3, Val Acc:  53.06%, Time: 0:01:15 *\nIter:    200, Train Loss:   0.61, Train Acc:  79.69%, Val Loss:   0.94, Val Acc:  69.88%, Time: 0:02:22 *\nIter:    300, Train Loss:   0.49, Train Acc:  85.16%, Val Loss:   0.63, Val Acc:  81.44%, Time: 0:03:29 *\nEpoch: 2\nIter:    400, Train Loss:   0.23, Train Acc:  92.97%, Val Loss:    0.6, Val Acc:  82.86%, Time: 0:04:36 *\nIter:    500, Train Loss:   0.27, Train Acc:  92.97%, Val Loss:   0.47, Val Acc:  86.72%, Time: 0:05:43 *\nIter:    600, Train Loss:   0.13, Train Acc:  98.44%, Val Loss:   0.43, Val Acc:  87.46%, Time: 0:06:50 *\nIter:    700, Train Loss:   0.24, Train Acc:  91.41%, Val Loss:   0.46, Val Acc:  87.12%, Time: 0:07:57\nEpoch: 3\nIter:    800, Train Loss:   0.11, Train Acc:  96.09%, Val Loss:   0.49, Val Acc:  87.02%, Time: 0:09:03\nIter:    900, Train Loss:   0.15, Train Acc:  96.09%, Val Loss:   0.55, Val Acc:  85.86%, Time: 0:10:10\nIter:   1000, Train Loss:   0.17, Train Acc:  96.09%, Val Loss:   0.43, Val Acc:  89.44%, Time: 0:11:18 *\nIter:   1100, Train Loss:   0.25, Train Acc:  93.75%, Val Loss:   0.42, Val Acc:  88.98%, Time: 0:12:25\nEpoch: 4\nIter:   1200, Train Loss:   0.14, Train Acc:  96.09%, Val Loss:   0.39, Val Acc:  89.82%, Time: 0:13:32 *\nIter:   1300, Train Loss:    0.2, Train Acc:  96.09%, Val Loss:   0.43, Val Acc:  88.68%, Time: 0:14:38\nIter:   1400, Train Loss:  0.012, Train Acc: 100.00%, Val Loss:   0.37, Val Acc:  90.58%, Time: 0:15:45 *\nIter:   1500, Train Loss:   0.15, Train Acc:  96.88%, Val Loss:   0.39, Val Acc:  90.58%, Time: 0:16:52\nEpoch: 5\nIter:   1600, Train Loss:  0.075, Train Acc:  97.66%, Val Loss:   0.41, Val Acc:  89.90%, Time: 0:17:59\nIter:   1700, Train Loss:  0.042, Train Acc:  98.44%, Val Loss:   0.41, Val Acc:  90.08%, Time: 0:19:06\nIter:   1800, Train Loss:   0.08, Train Acc:  97.66%, Val Loss:   0.38, Val Acc:  91.36%, Time: 0:20:13 *\nIter:   1900, Train Loss:  0.089, Train Acc:  98.44%, Val Loss:   0.39, Val Acc:  90.18%, Time: 0:21:20\nEpoch: 6\nIter:   2000, Train Loss:  0.092, Train Acc:  96.88%, Val Loss:   0.36, Val Acc:  91.42%, Time: 0:22:27 *\nIter:   2100, Train Loss:  0.062, Train Acc:  98.44%, Val Loss:   0.39, Val Acc:  90.56%, Time: 0:23:34\nIter:   2200, Train Loss:  0.053, Train Acc:  98.44%, Val Loss:   0.39, Val Acc:  90.02%, Time: 0:24:41\nIter:   2300, Train Loss:   0.12, Train Acc:  96.09%, Val Loss:   0.37, Val Acc:  90.84%, Time: 0:25:48\nEpoch: 7\nIter:   2400, Train Loss:  0.014, Train Acc: 100.00%, Val Loss:   0.41, Val Acc:  90.38%, Time: 0:26:55\nIter:   2500, Train Loss:   0.14, Train Acc:  96.88%, Val Loss:   0.37, Val Acc:  91.22%, Time: 0:28:01\nIter:   2600, Train Loss:   0.11, Train Acc:  96.88%, Val Loss:   0.43, Val Acc:  89.76%, Time: 0:29:08\nIter:   2700, Train Loss:  0.089, Train Acc:  97.66%, Val Loss:   0.37, Val Acc:  91.18%, Time: 0:30:15\nEpoch: 8\nIter:   2800, Train Loss: 0.0081, Train Acc: 100.00%, Val Loss:   0.44, Val Acc:  90.66%, Time: 0:31:22\nIter:   2900, Train Loss:  0.017, Train Acc: 100.00%, Val Loss:   0.44, Val Acc:  89.62%, Time: 0:32:29\nIter:   3000, Train Loss:  0.061, Train Acc:  96.88%, Val Loss:   0.43, Val Acc:  90.04%, Time: 0:33:36\nNo optimization for a long time, auto-stopping...\n```\n\n\u5728\u9a8c\u8bc1\u96c6\u4e0a\u7684\u6700\u4f73\u6548\u679c\u4e3a91.42%\uff0c\u7ecf\u8fc7\u4e868\u8f6e\u8fed\u4ee3\u505c\u6b62\uff0c\u901f\u5ea6\u76f8\u6bd4CNN\u6162\u5f88\u591a\u3002\n\n\u51c6\u786e\u7387\u548c\u8bef\u5dee\u5982\u56fe\u6240\u793a\uff1a\n\n![images](images/acc_loss_rnn.png)\n\n\n### \u6d4b\u8bd5\n\n\u8fd0\u884c `python run_rnn.py test` \u5728\u6d4b\u8bd5\u96c6\u4e0a\u8fdb\u884c\u6d4b\u8bd5\u3002\n\n```\nTesting...\nTest Loss:   0.21, Test Acc:  94.22%\nPrecision, Recall and F1-Score...\n             precision    recall  f1-score   support\n\n         \u4f53\u80b2       0.99      0.99      0.99      1000\n         \u8d22\u7ecf       0.91      0.99      0.95      1000\n         \u623f\u4ea7       1.00      1.00      1.00      1000\n         \u5bb6\u5c45       0.97      0.73      0.83      1000\n         \u6559\u80b2       0.91      0.92      0.91      1000\n         \u79d1\u6280       0.93      0.96      0.94      1000\n         \u65f6\u5c1a       0.89      0.97      0.93      1000\n         \u65f6\u653f       0.93      0.93      0.93      1000\n         \u6e38\u620f       0.95      0.97      0.96      1000\n         \u5a31\u4e50       0.97      0.96      0.97      1000\n\navg / total       0.94      0.94      0.94     10000\n\nConfusion Matrix...\n[[988   0   0   0   4   0   2   0   5   1]\n [  0 990   1   1   1   1   0   6   0   0]\n [  0   2 996   1   1   0   0   0   0   0]\n [  2  71   1 731  51  20  88  28   3   5]\n [  1   3   0   7 918  23   4  31   9   4]\n [  1   3   0   3   0 964   3   5  21   0]\n [  1   0   1   7   1   3 972   0   6   9]\n [  0  16   0   0  22  26   0 931   2   3]\n [  2   3   0   0   2   2  12   0 972   7]\n [  0   3   1   1   7   3  11   5   9 960]]\nTime usage: 0:00:33\n```\n\n\u5728\u6d4b\u8bd5\u96c6\u4e0a\u7684\u51c6\u786e\u7387\u8fbe\u5230\u4e8694.22%\uff0c\u4e14\u5404\u7c7b\u7684precision, recall\u548cf1-score\uff0c\u9664\u4e86\u5bb6\u5c45\u8fd9\u4e00\u7c7b\u522b\uff0c\u90fd\u8d85\u8fc7\u4e860.9\u3002\n\n\u4ece\u6df7\u6dc6\u77e9\u9635\u53ef\u4ee5\u770b\u51fa\u5206\u7c7b\u6548\u679c\u975e\u5e38\u4f18\u79c0\u3002\n\n\u5bf9\u6bd4\u4e24\u4e2a\u6a21\u578b\uff0c\u53ef\u89c1RNN\u9664\u4e86\u5728\u5bb6\u5c45\u5206\u7c7b\u7684\u8868\u73b0\u4e0d\u662f\u5f88\u7406\u60f3\uff0c\u5176\u4ed6\u51e0\u4e2a\u7c7b\u522b\u8f83CNN\u5dee\u522b\u4e0d\u5927\u3002\n\n\u8fd8\u53ef\u4ee5\u901a\u8fc7\u8fdb\u4e00\u6b65\u7684\u8c03\u8282\u53c2\u6570\uff0c\u6765\u8fbe\u5230\u66f4\u597d\u7684\u6548\u679c\u3002\n\n\n## \u9884\u6d4b\n\n\u4e3a\u65b9\u4fbf\u9884\u6d4b\uff0crepo \u4e2d `predict.py` \u63d0\u4f9b\u4e86 CNN \u6a21\u578b\u7684\u9884\u6d4b\u65b9\u6cd5\u3002\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": "zhanglei",
        "package_url": "https://pypi.org/project/zhanglei/",
        "platform": "",
        "project_url": "https://pypi.org/project/zhanglei/",
        "project_urls": {
            "Homepage": "https://github.com/pypa/sampleproject"
        },
        "release_url": "https://pypi.org/project/zhanglei/0.0.6/",
        "requires_dist": null,
        "requires_python": ">=3.6",
        "summary": "A small example package",
        "version": "0.0.6"
    },
    "last_serial": 5831911,
    "releases": {
        "0.0.2": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "78e460658aae23958d7cdedeed2e4955",
                    "sha256": "007f21f7b55f4ff4651d531e9975b5c736601a7739fca427645404cede04c7a4"
                },
                "downloads": -1,
                "filename": "zhanglei-0.0.2-py3-none-any.whl",
                "has_sig": false,
                "md5_digest": "78e460658aae23958d7cdedeed2e4955",
                "packagetype": "bdist_wheel",
                "python_version": "py3",
                "requires_python": ">=3.6",
                "size": 10170,
                "upload_time": "2019-09-15T12:36:26",
                "url": "https://files.pythonhosted.org/packages/a7/32/46caf85940dc45066f48754d9eaba2180506a43b8e7176064b2fa9ade1cc/zhanglei-0.0.2-py3-none-any.whl"
            },
            {
                "comment_text": "",
                "digests": {
                    "md5": "0a168d7e955a5ebcc201950ea118c351",
                    "sha256": "143a77d777670d3bc3037cabf2fc90d4d9ad14020abb90133b4f208874f254fa"
                },
                "downloads": -1,
                "filename": "zhanglei-0.0.2.tar.gz",
                "has_sig": false,
                "md5_digest": "0a168d7e955a5ebcc201950ea118c351",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": ">=3.6",
                "size": 10547,
                "upload_time": "2019-09-15T12:36:29",
                "url": "https://files.pythonhosted.org/packages/f4/d8/700571433a3ddfaf6345780f526dc73a62fe4e4f7155283ac6e5efc9e64c/zhanglei-0.0.2.tar.gz"
            }
        ],
        "0.0.3": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "422f0be1b3791bb44c6bdde92cbc6f9d",
                    "sha256": "1336eaa0bbd131dd82d4fc37115305e33f34bb73a9894d3b1fc95030c47e483c"
                },
                "downloads": -1,
                "filename": "zhanglei-0.0.3-py3-none-any.whl",
                "has_sig": false,
                "md5_digest": "422f0be1b3791bb44c6bdde92cbc6f9d",
                "packagetype": "bdist_wheel",
                "python_version": "py3",
                "requires_python": ">=3.6",
                "size": 11476,
                "upload_time": "2019-09-15T12:50:26",
                "url": "https://files.pythonhosted.org/packages/0d/45/d608d41acff47f42c727032596f7c1508bdddc4d492b6453db3a4d96907b/zhanglei-0.0.3-py3-none-any.whl"
            },
            {
                "comment_text": "",
                "digests": {
                    "md5": "f8fcb3133bf336698a48271d3b7af8bb",
                    "sha256": "92fecce212d375de5bb123ae6e22ca25a643d6c6b8492c4344a15622ba0c4bed"
                },
                "downloads": -1,
                "filename": "zhanglei-0.0.3.tar.gz",
                "has_sig": false,
                "md5_digest": "f8fcb3133bf336698a48271d3b7af8bb",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": ">=3.6",
                "size": 10549,
                "upload_time": "2019-09-15T12:50:30",
                "url": "https://files.pythonhosted.org/packages/5d/64/cb19c8e3f4cf186e6f62a64b5c05a97bf86477a8ade098dff99bb1178423/zhanglei-0.0.3.tar.gz"
            }
        ],
        "0.0.4": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "0d22b07bfe6e55f517bbaed50092422d",
                    "sha256": "bbfa30e330d13972c1a7f063ef37ce13d61be42de98c4886db211a5e0e0eef62"
                },
                "downloads": -1,
                "filename": "zhanglei-0.0.4-py3-none-any.whl",
                "has_sig": false,
                "md5_digest": "0d22b07bfe6e55f517bbaed50092422d",
                "packagetype": "bdist_wheel",
                "python_version": "py3",
                "requires_python": ">=3.6",
                "size": 22199,
                "upload_time": "2019-09-15T12:56:39",
                "url": "https://files.pythonhosted.org/packages/cc/82/82da349e53d719479b1a7b20e3e64d0de08b98fc1b46c9b7fa7247d18ad4/zhanglei-0.0.4-py3-none-any.whl"
            },
            {
                "comment_text": "",
                "digests": {
                    "md5": "03719ac1a7870eb9de0456e6fdeac800",
                    "sha256": "a94201a36d15ba6253200da51278d54580cada5f644975240e98596e23448e22"
                },
                "downloads": -1,
                "filename": "zhanglei-0.0.4.tar.gz",
                "has_sig": false,
                "md5_digest": "03719ac1a7870eb9de0456e6fdeac800",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": ">=3.6",
                "size": 18426,
                "upload_time": "2019-09-15T12:56:45",
                "url": "https://files.pythonhosted.org/packages/e9/db/f96ab5db5aa8257db5e8848e7defb170f3e6beeb39041e37ddddafe5d0c6/zhanglei-0.0.4.tar.gz"
            }
        ],
        "0.0.5": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "9cb2b9a1cedd29f13b8a290b01b55ee5",
                    "sha256": "99bf008ec56b4edfd5b126179730a19c6fb58f8120ddf1a53944a0025a8dc237"
                },
                "downloads": -1,
                "filename": "zhanglei-0.0.5-py3-none-any.whl",
                "has_sig": false,
                "md5_digest": "9cb2b9a1cedd29f13b8a290b01b55ee5",
                "packagetype": "bdist_wheel",
                "python_version": "py3",
                "requires_python": ">=3.6",
                "size": 22250,
                "upload_time": "2019-09-15T13:16:24",
                "url": "https://files.pythonhosted.org/packages/48/83/3a758e5e5c3e7f799eb6db93aa2af8ff589216fb230df024e570ff55dfdf/zhanglei-0.0.5-py3-none-any.whl"
            },
            {
                "comment_text": "",
                "digests": {
                    "md5": "1763c2523b94ae6373cc08da5aac1f4b",
                    "sha256": "6ae07dcabbacb0ff3a62562dd18ceaa1ef0473fe4535cc71e2f4c36298277501"
                },
                "downloads": -1,
                "filename": "zhanglei-0.0.5.tar.gz",
                "has_sig": false,
                "md5_digest": "1763c2523b94ae6373cc08da5aac1f4b",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": ">=3.6",
                "size": 18478,
                "upload_time": "2019-09-15T13:16:30",
                "url": "https://files.pythonhosted.org/packages/7e/55/d4bceea56a80d9233ba510eef2537268ac05ea83feeb08520a83e68b2c9b/zhanglei-0.0.5.tar.gz"
            }
        ],
        "0.0.6": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "fdd96d6822e4d1ec83830065662b5b7e",
                    "sha256": "674cdbe550b8d17da8fcac0a623ee31e1560f131539576a003f8b04ce556d34a"
                },
                "downloads": -1,
                "filename": "zhanglei-0.0.6-py3-none-any.whl",
                "has_sig": false,
                "md5_digest": "fdd96d6822e4d1ec83830065662b5b7e",
                "packagetype": "bdist_wheel",
                "python_version": "py3",
                "requires_python": ">=3.6",
                "size": 22244,
                "upload_time": "2019-09-15T13:23:59",
                "url": "https://files.pythonhosted.org/packages/6c/1f/db43535908abba64ba41eea1f8444ec17a8d02c8f53ba9d7df8c0aeec32e/zhanglei-0.0.6-py3-none-any.whl"
            },
            {
                "comment_text": "",
                "digests": {
                    "md5": "49605a5242be7d1f60f15b42bff1fcd8",
                    "sha256": "e84e15f613ee27bc8fa858bdc56a21554034e6f889d6a6e21e086eceec03ddbb"
                },
                "downloads": -1,
                "filename": "zhanglei-0.0.6.tar.gz",
                "has_sig": false,
                "md5_digest": "49605a5242be7d1f60f15b42bff1fcd8",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": ">=3.6",
                "size": 18471,
                "upload_time": "2019-09-15T13:24:01",
                "url": "https://files.pythonhosted.org/packages/0e/2e/eb00ac5f06a19d2a2f85c2c40ebcf0c4fe6c6366bb38da8464a03bbc1d2a/zhanglei-0.0.6.tar.gz"
            }
        ]
    },
    "urls": [
        {
            "comment_text": "",
            "digests": {
                "md5": "fdd96d6822e4d1ec83830065662b5b7e",
                "sha256": "674cdbe550b8d17da8fcac0a623ee31e1560f131539576a003f8b04ce556d34a"
            },
            "downloads": -1,
            "filename": "zhanglei-0.0.6-py3-none-any.whl",
            "has_sig": false,
            "md5_digest": "fdd96d6822e4d1ec83830065662b5b7e",
            "packagetype": "bdist_wheel",
            "python_version": "py3",
            "requires_python": ">=3.6",
            "size": 22244,
            "upload_time": "2019-09-15T13:23:59",
            "url": "https://files.pythonhosted.org/packages/6c/1f/db43535908abba64ba41eea1f8444ec17a8d02c8f53ba9d7df8c0aeec32e/zhanglei-0.0.6-py3-none-any.whl"
        },
        {
            "comment_text": "",
            "digests": {
                "md5": "49605a5242be7d1f60f15b42bff1fcd8",
                "sha256": "e84e15f613ee27bc8fa858bdc56a21554034e6f889d6a6e21e086eceec03ddbb"
            },
            "downloads": -1,
            "filename": "zhanglei-0.0.6.tar.gz",
            "has_sig": false,
            "md5_digest": "49605a5242be7d1f60f15b42bff1fcd8",
            "packagetype": "sdist",
            "python_version": "source",
            "requires_python": ">=3.6",
            "size": 18471,
            "upload_time": "2019-09-15T13:24:01",
            "url": "https://files.pythonhosted.org/packages/0e/2e/eb00ac5f06a19d2a2f85c2c40ebcf0c4fe6c6366bb38da8464a03bbc1d2a/zhanglei-0.0.6.tar.gz"
        }
    ]
}