[hint]Solution4a_NormalizationInsideNetwork.ipynb

{"cells":[{"cell_type":"markdown","source":["## Load Data"],"metadata":{"id":"ZW2T_03OXCZ4"},"id":"ZW2T_03OXCZ4"},{"cell_type":"code","execution_count":9,"id":"e175eeac","metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"e175eeac","executionInfo":{"status":"ok","timestamp":1669896765607,"user_tz":-420,"elapsed":716,"user":{"displayName":"Thanh Hùng","userId":"08335145449228425002"}},"outputId":"84f26921-b87c-4978-f9b4-1e4cb68cf91c"},"outputs":[{"output_type":"stream","name":"stdout","text":["(60000, 28, 28)\n","(60000,)\n","(10000, 28, 28)\n","(10000,)\n"]}],"source":["# fashion_mnist\n","import tensorflow as tf\n","import tensorflow.keras as keras\n","import numpy as np\n","tf.keras.utils.set_random_seed(1)\n","tf.config.experimental.enable_op_determinism()\n","\n","(X_train, y_train), (X_test, y_test) = keras.datasets.fashion_mnist.load_data()\n","print(X_train.shape)\n","print(y_train.shape)\n","print(X_test.shape)\n","print(y_test.shape)"]},{"cell_type":"markdown","source":["## Reshape Data"],"metadata":{"id":"f068e1ea"},"id":"f068e1ea"},{"cell_type":"code","execution_count":10,"id":"8aabf29a","metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"8aabf29a","executionInfo":{"status":"ok","timestamp":1669896765607,"user_tz":-420,"elapsed":3,"user":{"displayName":"Thanh Hùng","userId":"08335145449228425002"}},"outputId":"1e41b33d-8d29-429c-e359-9a6a1dd9c9f7"},"outputs":[{"output_type":"stream","name":"stdout","text":["(60000, 784)\n","(60000,)\n","(10000, 784)\n","(10000,)\n"]}],"source":["X_train = X_train.reshape(60000, -1)\n","X_test  = X_test.reshape(10000, -1)\n","\n","print(X_train.shape)\n","print(y_train.shape)\n","print(X_test.shape)\n","print(y_test.shape)"]},{"cell_type":"markdown","source":["## Normalization Inside Network"],"metadata":{"id":"c3167cdb"},"id":"c3167cdb"},{"cell_type":"code","execution_count":11,"id":"c46b65da","metadata":{"id":"c46b65da","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1669897657981,"user_tz":-420,"elapsed":892376,"user":{"displayName":"Thanh Hùng","userId":"08335145449228425002"}},"outputId":"69d4e208-c9b1-433c-c042-bf6c6e8c3252"},"outputs":[{"output_type":"stream","name":"stdout","text":["Epoch 1/300\n","118/118 - 5s - loss: 1.7121 - accuracy: 0.5625 - val_loss: 2.3062 - val_accuracy: 0.1000 - 5s/epoch - 44ms/step\n","Epoch 2/300\n","118/118 - 3s - loss: 1.1582 - accuracy: 0.7408 - val_loss: 2.2837 - val_accuracy: 0.1650 - 3s/epoch - 25ms/step\n","Epoch 3/300\n","118/118 - 3s - loss: 0.9393 - accuracy: 0.7675 - val_loss: 1.9771 - val_accuracy: 0.6084 - 3s/epoch - 25ms/step\n","Epoch 4/300\n","118/118 - 3s - loss: 0.8110 - accuracy: 0.7842 - val_loss: 1.2107 - val_accuracy: 0.7783 - 3s/epoch - 25ms/step\n","Epoch 5/300\n","118/118 - 3s - loss: 0.7266 - accuracy: 0.7972 - val_loss: 0.8069 - val_accuracy: 0.7934 - 3s/epoch - 25ms/step\n","Epoch 6/300\n","118/118 - 3s - loss: 0.6662 - accuracy: 0.8086 - val_loss: 0.6778 - val_accuracy: 0.8061 - 3s/epoch - 25ms/step\n","Epoch 7/300\n","118/118 - 3s - loss: 0.6218 - accuracy: 0.8171 - val_loss: 0.6244 - val_accuracy: 0.8109 - 3s/epoch - 25ms/step\n","Epoch 8/300\n","118/118 - 3s - loss: 0.5876 - accuracy: 0.8232 - val_loss: 0.5921 - val_accuracy: 0.8166 - 3s/epoch - 25ms/step\n","Epoch 9/300\n","118/118 - 3s - loss: 0.5604 - accuracy: 0.8289 - val_loss: 0.5688 - val_accuracy: 0.8220 - 3s/epoch - 25ms/step\n","Epoch 10/300\n","118/118 - 3s - loss: 0.5393 - accuracy: 0.8320 - val_loss: 0.5511 - val_accuracy: 0.8263 - 3s/epoch - 25ms/step\n","Epoch 11/300\n","118/118 - 3s - loss: 0.5206 - accuracy: 0.8366 - val_loss: 0.5382 - val_accuracy: 0.8256 - 3s/epoch - 25ms/step\n","Epoch 12/300\n","118/118 - 3s - loss: 0.5050 - accuracy: 0.8401 - val_loss: 0.5285 - val_accuracy: 0.8260 - 3s/epoch - 25ms/step\n","Epoch 13/300\n","118/118 - 3s - loss: 0.4918 - accuracy: 0.8423 - val_loss: 0.5122 - val_accuracy: 0.8293 - 3s/epoch - 25ms/step\n","Epoch 14/300\n","118/118 - 3s - loss: 0.4801 - accuracy: 0.8453 - val_loss: 0.5025 - val_accuracy: 0.8305 - 3s/epoch - 25ms/step\n","Epoch 15/300\n","118/118 - 3s - loss: 0.4707 - accuracy: 0.8459 - val_loss: 0.4968 - val_accuracy: 0.8309 - 3s/epoch - 25ms/step\n","Epoch 16/300\n","118/118 - 3s - loss: 0.4613 - accuracy: 0.8490 - val_loss: 0.4888 - val_accuracy: 0.8332 - 3s/epoch - 25ms/step\n","Epoch 17/300\n","118/118 - 3s - loss: 0.4529 - accuracy: 0.8502 - val_loss: 0.4870 - val_accuracy: 0.8331 - 3s/epoch - 25ms/step\n","Epoch 18/300\n","118/118 - 3s - loss: 0.4458 - accuracy: 0.8528 - val_loss: 0.4755 - val_accuracy: 0.8377 - 3s/epoch - 25ms/step\n","Epoch 19/300\n","118/118 - 3s - loss: 0.4392 - accuracy: 0.8538 - val_loss: 0.4708 - val_accuracy: 0.8353 - 3s/epoch - 25ms/step\n","Epoch 20/300\n","118/118 - 3s - loss: 0.4333 - accuracy: 0.8553 - val_loss: 0.4640 - val_accuracy: 0.8401 - 3s/epoch - 25ms/step\n","Epoch 21/300\n","118/118 - 3s - loss: 0.4272 - accuracy: 0.8571 - val_loss: 0.4608 - val_accuracy: 0.8373 - 3s/epoch - 25ms/step\n","Epoch 22/300\n","118/118 - 3s - loss: 0.4221 - accuracy: 0.8587 - val_loss: 0.4572 - val_accuracy: 0.8401 - 3s/epoch - 25ms/step\n","Epoch 23/300\n","118/118 - 3s - loss: 0.4167 - accuracy: 0.8594 - val_loss: 0.4549 - val_accuracy: 0.8406 - 3s/epoch - 25ms/step\n","Epoch 24/300\n","118/118 - 3s - loss: 0.4125 - accuracy: 0.8614 - val_loss: 0.4474 - val_accuracy: 0.8427 - 3s/epoch - 25ms/step\n","Epoch 25/300\n","118/118 - 3s - loss: 0.4083 - accuracy: 0.8616 - val_loss: 0.4455 - val_accuracy: 0.8441 - 3s/epoch - 25ms/step\n","Epoch 26/300\n","118/118 - 3s - loss: 0.4035 - accuracy: 0.8633 - val_loss: 0.4422 - val_accuracy: 0.8451 - 3s/epoch - 24ms/step\n","Epoch 27/300\n","118/118 - 3s - loss: 0.3995 - accuracy: 0.8640 - val_loss: 0.4411 - val_accuracy: 0.8439 - 3s/epoch - 25ms/step\n","Epoch 28/300\n","118/118 - 3s - loss: 0.3953 - accuracy: 0.8662 - val_loss: 0.4467 - val_accuracy: 0.8428 - 3s/epoch - 25ms/step\n","Epoch 29/300\n","118/118 - 3s - loss: 0.3920 - accuracy: 0.8659 - val_loss: 0.4335 - val_accuracy: 0.8444 - 3s/epoch - 25ms/step\n","Epoch 30/300\n","118/118 - 3s - loss: 0.3891 - accuracy: 0.8681 - val_loss: 0.4326 - val_accuracy: 0.8439 - 3s/epoch - 25ms/step\n","Epoch 31/300\n","118/118 - 3s - loss: 0.3848 - accuracy: 0.8685 - val_loss: 0.4290 - val_accuracy: 0.8457 - 3s/epoch - 25ms/step\n","Epoch 32/300\n","118/118 - 3s - loss: 0.3817 - accuracy: 0.8694 - val_loss: 0.4245 - val_accuracy: 0.8492 - 3s/epoch - 25ms/step\n","Epoch 33/300\n","118/118 - 3s - loss: 0.3778 - accuracy: 0.8713 - val_loss: 0.4239 - val_accuracy: 0.8479 - 3s/epoch - 25ms/step\n","Epoch 34/300\n","118/118 - 3s - loss: 0.3742 - accuracy: 0.8732 - val_loss: 0.4281 - val_accuracy: 0.8466 - 3s/epoch - 25ms/step\n","Epoch 35/300\n","118/118 - 3s - loss: 0.3731 - accuracy: 0.8727 - val_loss: 0.4212 - val_accuracy: 0.8496 - 3s/epoch - 25ms/step\n","Epoch 36/300\n","118/118 - 3s - loss: 0.3690 - accuracy: 0.8734 - val_loss: 0.4184 - val_accuracy: 0.8498 - 3s/epoch - 25ms/step\n","Epoch 37/300\n","118/118 - 3s - loss: 0.3671 - accuracy: 0.8747 - val_loss: 0.4211 - val_accuracy: 0.8468 - 3s/epoch - 25ms/step\n","Epoch 38/300\n","118/118 - 3s - loss: 0.3641 - accuracy: 0.8754 - val_loss: 0.4170 - val_accuracy: 0.8501 - 3s/epoch - 25ms/step\n","Epoch 39/300\n","118/118 - 3s - loss: 0.3605 - accuracy: 0.8763 - val_loss: 0.4154 - val_accuracy: 0.8518 - 3s/epoch - 25ms/step\n","Epoch 40/300\n","118/118 - 3s - loss: 0.3589 - accuracy: 0.8771 - val_loss: 0.4191 - val_accuracy: 0.8505 - 3s/epoch - 25ms/step\n","Epoch 41/300\n","118/118 - 3s - loss: 0.3561 - accuracy: 0.8776 - val_loss: 0.4130 - val_accuracy: 0.8536 - 3s/epoch - 25ms/step\n","Epoch 42/300\n","118/118 - 3s - loss: 0.3536 - accuracy: 0.8785 - val_loss: 0.4387 - val_accuracy: 0.8413 - 3s/epoch - 25ms/step\n","Epoch 43/300\n","118/118 - 3s - loss: 0.3514 - accuracy: 0.8793 - val_loss: 0.4188 - val_accuracy: 0.8520 - 3s/epoch - 25ms/step\n","Epoch 44/300\n","118/118 - 4s - loss: 0.3491 - accuracy: 0.8799 - val_loss: 0.4027 - val_accuracy: 0.8558 - 4s/epoch - 30ms/step\n","Epoch 45/300\n","118/118 - 3s - loss: 0.3459 - accuracy: 0.8811 - val_loss: 0.4036 - val_accuracy: 0.8553 - 3s/epoch - 25ms/step\n","Epoch 46/300\n","118/118 - 3s - loss: 0.3442 - accuracy: 0.8812 - val_loss: 0.4284 - val_accuracy: 0.8480 - 3s/epoch - 25ms/step\n","Epoch 47/300\n","118/118 - 3s - loss: 0.3423 - accuracy: 0.8821 - val_loss: 0.4096 - val_accuracy: 0.8554 - 3s/epoch - 25ms/step\n","Epoch 48/300\n","118/118 - 3s - loss: 0.3402 - accuracy: 0.8821 - val_loss: 0.3986 - val_accuracy: 0.8571 - 3s/epoch - 25ms/step\n","Epoch 49/300\n","118/118 - 3s - loss: 0.3381 - accuracy: 0.8824 - val_loss: 0.3991 - val_accuracy: 0.8548 - 3s/epoch - 25ms/step\n","Epoch 50/300\n","118/118 - 3s - loss: 0.3353 - accuracy: 0.8848 - val_loss: 0.4003 - val_accuracy: 0.8555 - 3s/epoch - 25ms/step\n","Epoch 51/300\n","118/118 - 3s - loss: 0.3343 - accuracy: 0.8849 - val_loss: 0.3965 - val_accuracy: 0.8594 - 3s/epoch - 25ms/step\n","Epoch 52/300\n","118/118 - 3s - loss: 0.3320 - accuracy: 0.8857 - val_loss: 0.3928 - val_accuracy: 0.8615 - 3s/epoch - 25ms/step\n","Epoch 53/300\n","118/118 - 3s - loss: 0.3303 - accuracy: 0.8850 - val_loss: 0.3930 - val_accuracy: 0.8598 - 3s/epoch - 25ms/step\n","Epoch 54/300\n","118/118 - 3s - loss: 0.3285 - accuracy: 0.8854 - val_loss: 0.4019 - val_accuracy: 0.8558 - 3s/epoch - 25ms/step\n","Epoch 55/300\n","118/118 - 3s - loss: 0.3273 - accuracy: 0.8868 - val_loss: 0.4178 - val_accuracy: 0.8498 - 3s/epoch - 25ms/step\n","Epoch 56/300\n","118/118 - 3s - loss: 0.3248 - accuracy: 0.8873 - val_loss: 0.3890 - val_accuracy: 0.8613 - 3s/epoch - 25ms/step\n","Epoch 57/300\n","118/118 - 3s - loss: 0.3226 - accuracy: 0.8884 - val_loss: 0.3859 - val_accuracy: 0.8638 - 3s/epoch - 25ms/step\n","Epoch 58/300\n","118/118 - 3s - loss: 0.3210 - accuracy: 0.8869 - val_loss: 0.4028 - val_accuracy: 0.8572 - 3s/epoch - 25ms/step\n","Epoch 59/300\n","118/118 - 3s - loss: 0.3186 - accuracy: 0.8892 - val_loss: 0.3905 - val_accuracy: 0.8599 - 3s/epoch - 25ms/step\n","Epoch 60/300\n","118/118 - 3s - loss: 0.3171 - accuracy: 0.8890 - val_loss: 0.4267 - val_accuracy: 0.8486 - 3s/epoch - 25ms/step\n","Epoch 61/300\n","118/118 - 3s - loss: 0.3162 - accuracy: 0.8904 - val_loss: 0.3888 - val_accuracy: 0.8603 - 3s/epoch - 25ms/step\n","Epoch 62/300\n","118/118 - 3s - loss: 0.3137 - accuracy: 0.8909 - val_loss: 0.4139 - val_accuracy: 0.8509 - 3s/epoch - 25ms/step\n","Epoch 63/300\n","118/118 - 3s - loss: 0.3126 - accuracy: 0.8919 - val_loss: 0.3937 - val_accuracy: 0.8600 - 3s/epoch - 25ms/step\n","Epoch 64/300\n","118/118 - 3s - loss: 0.3099 - accuracy: 0.8915 - val_loss: 0.3897 - val_accuracy: 0.8610 - 3s/epoch - 25ms/step\n","Epoch 65/300\n","118/118 - 3s - loss: 0.3097 - accuracy: 0.8921 - val_loss: 0.3983 - val_accuracy: 0.8559 - 3s/epoch - 25ms/step\n","Epoch 66/300\n","118/118 - 3s - loss: 0.3073 - accuracy: 0.8921 - val_loss: 0.4068 - val_accuracy: 0.8540 - 3s/epoch - 25ms/step\n","Epoch 67/300\n","118/118 - 3s - loss: 0.3048 - accuracy: 0.8931 - val_loss: 0.4029 - val_accuracy: 0.8569 - 3s/epoch - 25ms/step\n","Epoch 68/300\n","118/118 - 3s - loss: 0.3034 - accuracy: 0.8936 - val_loss: 0.3857 - val_accuracy: 0.8615 - 3s/epoch - 25ms/step\n","Epoch 69/300\n","118/118 - 3s - loss: 0.3019 - accuracy: 0.8953 - val_loss: 0.3846 - val_accuracy: 0.8625 - 3s/epoch - 25ms/step\n","Epoch 70/300\n","118/118 - 3s - loss: 0.3003 - accuracy: 0.8946 - val_loss: 0.3954 - val_accuracy: 0.8578 - 3s/epoch - 25ms/step\n","Epoch 71/300\n","118/118 - 3s - loss: 0.2997 - accuracy: 0.8955 - val_loss: 0.3840 - val_accuracy: 0.8637 - 3s/epoch - 25ms/step\n","Epoch 72/300\n","118/118 - 3s - loss: 0.2976 - accuracy: 0.8959 - val_loss: 0.4029 - val_accuracy: 0.8553 - 3s/epoch - 24ms/step\n","Epoch 73/300\n","118/118 - 3s - loss: 0.2965 - accuracy: 0.8957 - val_loss: 0.4334 - val_accuracy: 0.8455 - 3s/epoch - 25ms/step\n","Epoch 74/300\n","118/118 - 3s - loss: 0.2961 - accuracy: 0.8960 - val_loss: 0.4269 - val_accuracy: 0.8480 - 3s/epoch - 25ms/step\n","Epoch 75/300\n","118/118 - 3s - loss: 0.2949 - accuracy: 0.8965 - val_loss: 0.3849 - val_accuracy: 0.8631 - 3s/epoch - 25ms/step\n","Epoch 76/300\n","118/118 - 3s - loss: 0.2917 - accuracy: 0.8976 - val_loss: 0.3838 - val_accuracy: 0.8637 - 3s/epoch - 25ms/step\n","Epoch 77/300\n","118/118 - 3s - loss: 0.2894 - accuracy: 0.8982 - val_loss: 0.3814 - val_accuracy: 0.8639 - 3s/epoch - 25ms/step\n","Epoch 78/300\n","118/118 - 3s - loss: 0.2886 - accuracy: 0.8988 - val_loss: 0.3828 - val_accuracy: 0.8649 - 3s/epoch - 25ms/step\n","Epoch 79/300\n","118/118 - 3s - loss: 0.2885 - accuracy: 0.8992 - val_loss: 0.3835 - val_accuracy: 0.8609 - 3s/epoch - 25ms/step\n","Epoch 80/300\n","118/118 - 3s - loss: 0.2849 - accuracy: 0.8996 - val_loss: 0.3790 - val_accuracy: 0.8648 - 3s/epoch - 25ms/step\n","Epoch 81/300\n","118/118 - 3s - loss: 0.2855 - accuracy: 0.8994 - val_loss: 0.3920 - val_accuracy: 0.8607 - 3s/epoch - 26ms/step\n","Epoch 82/300\n","118/118 - 3s - loss: 0.2838 - accuracy: 0.9003 - val_loss: 0.3819 - val_accuracy: 0.8656 - 3s/epoch - 25ms/step\n","Epoch 83/300\n","118/118 - 3s - loss: 0.2816 - accuracy: 0.9012 - val_loss: 0.3925 - val_accuracy: 0.8609 - 3s/epoch - 25ms/step\n","Epoch 84/300\n","118/118 - 3s - loss: 0.2808 - accuracy: 0.9015 - val_loss: 0.3742 - val_accuracy: 0.8692 - 3s/epoch - 25ms/step\n","Epoch 85/300\n","118/118 - 3s - loss: 0.2788 - accuracy: 0.9012 - val_loss: 0.3857 - val_accuracy: 0.8621 - 3s/epoch - 25ms/step\n","Epoch 86/300\n","118/118 - 3s - loss: 0.2781 - accuracy: 0.9020 - val_loss: 0.3827 - val_accuracy: 0.8677 - 3s/epoch - 26ms/step\n","Epoch 87/300\n","118/118 - 3s - loss: 0.2752 - accuracy: 0.9024 - val_loss: 0.3837 - val_accuracy: 0.8627 - 3s/epoch - 25ms/step\n","Epoch 88/300\n","118/118 - 3s - loss: 0.2740 - accuracy: 0.9034 - val_loss: 0.3937 - val_accuracy: 0.8582 - 3s/epoch - 25ms/step\n","Epoch 89/300\n","118/118 - 3s - loss: 0.2744 - accuracy: 0.9030 - val_loss: 0.4036 - val_accuracy: 0.8556 - 3s/epoch - 25ms/step\n","Epoch 90/300\n","118/118 - 3s - loss: 0.2732 - accuracy: 0.9031 - val_loss: 0.3745 - val_accuracy: 0.8666 - 3s/epoch - 25ms/step\n","Epoch 91/300\n","118/118 - 3s - loss: 0.2703 - accuracy: 0.9046 - val_loss: 0.5072 - val_accuracy: 0.8244 - 3s/epoch - 25ms/step\n","Epoch 92/300\n","118/118 - 3s - loss: 0.2697 - accuracy: 0.9046 - val_loss: 0.3772 - val_accuracy: 0.8686 - 3s/epoch - 25ms/step\n","Epoch 93/300\n","118/118 - 3s - loss: 0.2689 - accuracy: 0.9051 - val_loss: 0.3886 - val_accuracy: 0.8628 - 3s/epoch - 25ms/step\n","Epoch 94/300\n","118/118 - 3s - loss: 0.2680 - accuracy: 0.9063 - val_loss: 0.3921 - val_accuracy: 0.8620 - 3s/epoch - 25ms/step\n","Epoch 95/300\n","118/118 - 3s - loss: 0.2652 - accuracy: 0.9065 - val_loss: 0.3762 - val_accuracy: 0.8650 - 3s/epoch - 25ms/step\n","Epoch 96/300\n","118/118 - 3s - loss: 0.2643 - accuracy: 0.9069 - val_loss: 0.3938 - val_accuracy: 0.8588 - 3s/epoch - 25ms/step\n","Epoch 97/300\n","118/118 - 3s - loss: 0.2640 - accuracy: 0.9074 - val_loss: 0.3780 - val_accuracy: 0.8689 - 3s/epoch - 25ms/step\n","Epoch 98/300\n","118/118 - 3s - loss: 0.2633 - accuracy: 0.9071 - val_loss: 0.4125 - val_accuracy: 0.8527 - 3s/epoch - 25ms/step\n","Epoch 99/300\n","118/118 - 3s - loss: 0.2615 - accuracy: 0.9080 - val_loss: 0.3762 - val_accuracy: 0.8703 - 3s/epoch - 25ms/step\n","Epoch 100/300\n","118/118 - 3s - loss: 0.2610 - accuracy: 0.9078 - val_loss: 0.3986 - val_accuracy: 0.8624 - 3s/epoch - 25ms/step\n","Epoch 101/300\n","118/118 - 3s - loss: 0.2590 - accuracy: 0.9090 - val_loss: 0.3935 - val_accuracy: 0.8601 - 3s/epoch - 25ms/step\n","Epoch 102/300\n","118/118 - 3s - loss: 0.2569 - accuracy: 0.9093 - val_loss: 0.4033 - val_accuracy: 0.8549 - 3s/epoch - 25ms/step\n","Epoch 103/300\n","118/118 - 3s - loss: 0.2559 - accuracy: 0.9096 - val_loss: 0.4138 - val_accuracy: 0.8516 - 3s/epoch - 26ms/step\n","Epoch 104/300\n","118/118 - 3s - loss: 0.2554 - accuracy: 0.9091 - val_loss: 0.3694 - val_accuracy: 0.8698 - 3s/epoch - 25ms/step\n","Epoch 105/300\n","118/118 - 3s - loss: 0.2547 - accuracy: 0.9103 - val_loss: 0.3879 - val_accuracy: 0.8666 - 3s/epoch - 26ms/step\n","Epoch 106/300\n","118/118 - 3s - loss: 0.2537 - accuracy: 0.9103 - val_loss: 0.4266 - val_accuracy: 0.8522 - 3s/epoch - 26ms/step\n","Epoch 107/300\n","118/118 - 3s - loss: 0.2516 - accuracy: 0.9107 - val_loss: 0.3717 - val_accuracy: 0.8697 - 3s/epoch - 25ms/step\n","Epoch 108/300\n","118/118 - 3s - loss: 0.2502 - accuracy: 0.9114 - val_loss: 0.3657 - val_accuracy: 0.8725 - 3s/epoch - 26ms/step\n","Epoch 109/300\n","118/118 - 3s - loss: 0.2497 - accuracy: 0.9115 - val_loss: 0.3710 - val_accuracy: 0.8710 - 3s/epoch - 26ms/step\n","Epoch 110/300\n","118/118 - 3s - loss: 0.2483 - accuracy: 0.9118 - val_loss: 0.4144 - val_accuracy: 0.8565 - 3s/epoch - 26ms/step\n","Epoch 111/300\n","118/118 - 3s - loss: 0.2484 - accuracy: 0.9117 - val_loss: 0.3679 - val_accuracy: 0.8715 - 3s/epoch - 26ms/step\n","Epoch 112/300\n","118/118 - 3s - loss: 0.2463 - accuracy: 0.9132 - val_loss: 0.3771 - val_accuracy: 0.8686 - 3s/epoch - 26ms/step\n","Epoch 113/300\n","118/118 - 3s - loss: 0.2449 - accuracy: 0.9137 - val_loss: 0.3928 - val_accuracy: 0.8599 - 3s/epoch - 26ms/step\n","Epoch 114/300\n","118/118 - 3s - loss: 0.2437 - accuracy: 0.9144 - val_loss: 0.3907 - val_accuracy: 0.8609 - 3s/epoch - 25ms/step\n","Epoch 115/300\n","118/118 - 3s - loss: 0.2427 - accuracy: 0.9145 - val_loss: 0.3920 - val_accuracy: 0.8583 - 3s/epoch - 25ms/step\n","Epoch 116/300\n","118/118 - 3s - loss: 0.2409 - accuracy: 0.9137 - val_loss: 0.3791 - val_accuracy: 0.8683 - 3s/epoch - 25ms/step\n","Epoch 117/300\n","118/118 - 3s - loss: 0.2411 - accuracy: 0.9140 - val_loss: 0.4017 - val_accuracy: 0.8548 - 3s/epoch - 25ms/step\n","Epoch 118/300\n","118/118 - 3s - loss: 0.2407 - accuracy: 0.9146 - val_loss: 0.3680 - val_accuracy: 0.8684 - 3s/epoch - 25ms/step\n","Epoch 119/300\n","118/118 - 3s - loss: 0.2369 - accuracy: 0.9166 - val_loss: 0.3808 - val_accuracy: 0.8669 - 3s/epoch - 26ms/step\n","Epoch 120/300\n","118/118 - 4s - loss: 0.2382 - accuracy: 0.9159 - val_loss: 0.4055 - val_accuracy: 0.8580 - 4s/epoch - 31ms/step\n","Epoch 121/300\n","118/118 - 3s - loss: 0.2364 - accuracy: 0.9162 - val_loss: 0.3678 - val_accuracy: 0.8701 - 3s/epoch - 25ms/step\n","Epoch 122/300\n","118/118 - 3s - loss: 0.2352 - accuracy: 0.9175 - val_loss: 0.3719 - val_accuracy: 0.8690 - 3s/epoch - 25ms/step\n","Epoch 123/300\n","118/118 - 3s - loss: 0.2344 - accuracy: 0.9170 - val_loss: 0.3831 - val_accuracy: 0.8671 - 3s/epoch - 25ms/step\n","Epoch 124/300\n","118/118 - 3s - loss: 0.2329 - accuracy: 0.9169 - val_loss: 0.3731 - val_accuracy: 0.8684 - 3s/epoch - 25ms/step\n","Epoch 125/300\n","118/118 - 3s - loss: 0.2335 - accuracy: 0.9179 - val_loss: 0.3791 - val_accuracy: 0.8689 - 3s/epoch - 25ms/step\n","Epoch 126/300\n","118/118 - 3s - loss: 0.2326 - accuracy: 0.9176 - val_loss: 0.4082 - val_accuracy: 0.8580 - 3s/epoch - 25ms/step\n","Epoch 127/300\n","118/118 - 3s - loss: 0.2296 - accuracy: 0.9200 - val_loss: 0.3778 - val_accuracy: 0.8655 - 3s/epoch - 25ms/step\n","Epoch 128/300\n","118/118 - 3s - loss: 0.2295 - accuracy: 0.9181 - val_loss: 0.4018 - val_accuracy: 0.8626 - 3s/epoch - 25ms/step\n","Epoch 129/300\n","118/118 - 3s - loss: 0.2294 - accuracy: 0.9190 - val_loss: 0.3854 - val_accuracy: 0.8616 - 3s/epoch - 25ms/step\n","Epoch 130/300\n","118/118 - 3s - loss: 0.2276 - accuracy: 0.9186 - val_loss: 0.3716 - val_accuracy: 0.8688 - 3s/epoch - 25ms/step\n","Epoch 131/300\n","118/118 - 3s - loss: 0.2252 - accuracy: 0.9200 - val_loss: 0.3977 - val_accuracy: 0.8622 - 3s/epoch - 25ms/step\n","Epoch 132/300\n","118/118 - 3s - loss: 0.2248 - accuracy: 0.9208 - val_loss: 0.3830 - val_accuracy: 0.8657 - 3s/epoch - 25ms/step\n","Epoch 133/300\n","118/118 - 3s - loss: 0.2229 - accuracy: 0.9209 - val_loss: 0.4070 - val_accuracy: 0.8563 - 3s/epoch - 25ms/step\n","Epoch 134/300\n","118/118 - 3s - loss: 0.2234 - accuracy: 0.9221 - val_loss: 0.4083 - val_accuracy: 0.8577 - 3s/epoch - 25ms/step\n","Epoch 135/300\n","118/118 - 3s - loss: 0.2213 - accuracy: 0.9211 - val_loss: 0.3671 - val_accuracy: 0.8740 - 3s/epoch - 25ms/step\n","Epoch 136/300\n","118/118 - 3s - loss: 0.2203 - accuracy: 0.9220 - val_loss: 0.3759 - val_accuracy: 0.8668 - 3s/epoch - 25ms/step\n","Epoch 137/300\n","118/118 - 3s - loss: 0.2200 - accuracy: 0.9218 - val_loss: 0.3731 - val_accuracy: 0.8718 - 3s/epoch - 25ms/step\n","Epoch 138/300\n","118/118 - 3s - loss: 0.2200 - accuracy: 0.9225 - val_loss: 0.4289 - val_accuracy: 0.8482 - 3s/epoch - 25ms/step\n","Epoch 139/300\n","118/118 - 3s - loss: 0.2194 - accuracy: 0.9228 - val_loss: 0.3863 - val_accuracy: 0.8661 - 3s/epoch - 25ms/step\n","Epoch 140/300\n","118/118 - 3s - loss: 0.2183 - accuracy: 0.9230 - val_loss: 0.3823 - val_accuracy: 0.8681 - 3s/epoch - 25ms/step\n","Epoch 141/300\n","118/118 - 3s - loss: 0.2162 - accuracy: 0.9233 - val_loss: 0.4055 - val_accuracy: 0.8612 - 3s/epoch - 25ms/step\n","Epoch 142/300\n","118/118 - 3s - loss: 0.2153 - accuracy: 0.9244 - val_loss: 0.3915 - val_accuracy: 0.8655 - 3s/epoch - 25ms/step\n","Epoch 143/300\n","118/118 - 3s - loss: 0.2145 - accuracy: 0.9244 - val_loss: 0.4094 - val_accuracy: 0.8564 - 3s/epoch - 25ms/step\n","Epoch 144/300\n","118/118 - 3s - loss: 0.2147 - accuracy: 0.9239 - val_loss: 0.3864 - val_accuracy: 0.8630 - 3s/epoch - 25ms/step\n","Epoch 145/300\n","118/118 - 3s - loss: 0.2128 - accuracy: 0.9252 - val_loss: 0.3858 - val_accuracy: 0.8669 - 3s/epoch - 25ms/step\n","Epoch 146/300\n","118/118 - 3s - loss: 0.2113 - accuracy: 0.9258 - val_loss: 0.3713 - val_accuracy: 0.8736 - 3s/epoch - 25ms/step\n","Epoch 147/300\n","118/118 - 3s - loss: 0.2116 - accuracy: 0.9254 - val_loss: 0.3896 - val_accuracy: 0.8669 - 3s/epoch - 25ms/step\n","Epoch 148/300\n","118/118 - 3s - loss: 0.2104 - accuracy: 0.9258 - val_loss: 0.4065 - val_accuracy: 0.8634 - 3s/epoch - 25ms/step\n","Epoch 149/300\n","118/118 - 3s - loss: 0.2105 - accuracy: 0.9260 - val_loss: 0.3854 - val_accuracy: 0.8667 - 3s/epoch - 25ms/step\n","Epoch 150/300\n","118/118 - 3s - loss: 0.2086 - accuracy: 0.9267 - val_loss: 0.4052 - val_accuracy: 0.8628 - 3s/epoch - 25ms/step\n","Epoch 151/300\n","118/118 - 3s - loss: 0.2083 - accuracy: 0.9268 - val_loss: 0.4157 - val_accuracy: 0.8593 - 3s/epoch - 25ms/step\n","Epoch 152/300\n","118/118 - 3s - loss: 0.2055 - accuracy: 0.9274 - val_loss: 0.4079 - val_accuracy: 0.8576 - 3s/epoch - 25ms/step\n","Epoch 153/300\n","118/118 - 3s - loss: 0.2045 - accuracy: 0.9286 - val_loss: 0.3962 - val_accuracy: 0.8634 - 3s/epoch - 25ms/step\n","Epoch 154/300\n","118/118 - 3s - loss: 0.2041 - accuracy: 0.9279 - val_loss: 0.4221 - val_accuracy: 0.8553 - 3s/epoch - 25ms/step\n","Epoch 155/300\n","118/118 - 3s - loss: 0.2035 - accuracy: 0.9278 - val_loss: 0.3899 - val_accuracy: 0.8672 - 3s/epoch - 25ms/step\n","Epoch 156/300\n","118/118 - 3s - loss: 0.2030 - accuracy: 0.9291 - val_loss: 0.4172 - val_accuracy: 0.8535 - 3s/epoch - 25ms/step\n","Epoch 157/300\n","118/118 - 3s - loss: 0.2019 - accuracy: 0.9289 - val_loss: 0.3849 - val_accuracy: 0.8671 - 3s/epoch - 25ms/step\n","Epoch 158/300\n","118/118 - 3s - loss: 0.2012 - accuracy: 0.9294 - val_loss: 0.3901 - val_accuracy: 0.8687 - 3s/epoch - 25ms/step\n","Epoch 159/300\n","118/118 - 3s - loss: 0.2001 - accuracy: 0.9298 - val_loss: 0.4551 - val_accuracy: 0.8431 - 3s/epoch - 25ms/step\n","Epoch 160/300\n","118/118 - 3s - loss: 0.1994 - accuracy: 0.9305 - val_loss: 0.3918 - val_accuracy: 0.8654 - 3s/epoch - 25ms/step\n","Epoch 161/300\n","118/118 - 3s - loss: 0.1989 - accuracy: 0.9306 - val_loss: 0.3800 - val_accuracy: 0.8725 - 3s/epoch - 25ms/step\n","Epoch 162/300\n","118/118 - 3s - loss: 0.1979 - accuracy: 0.9307 - val_loss: 0.3912 - val_accuracy: 0.8641 - 3s/epoch - 25ms/step\n","Epoch 163/300\n","118/118 - 3s - loss: 0.1966 - accuracy: 0.9312 - val_loss: 0.3851 - val_accuracy: 0.8665 - 3s/epoch - 25ms/step\n","Epoch 164/300\n","118/118 - 3s - loss: 0.1953 - accuracy: 0.9316 - val_loss: 0.4053 - val_accuracy: 0.8645 - 3s/epoch - 25ms/step\n","Epoch 165/300\n","118/118 - 3s - loss: 0.1955 - accuracy: 0.9314 - val_loss: 0.3933 - val_accuracy: 0.8674 - 3s/epoch - 25ms/step\n","Epoch 166/300\n","118/118 - 3s - loss: 0.1939 - accuracy: 0.9318 - val_loss: 0.3754 - val_accuracy: 0.8717 - 3s/epoch - 25ms/step\n","Epoch 167/300\n","118/118 - 3s - loss: 0.1925 - accuracy: 0.9319 - val_loss: 0.3723 - val_accuracy: 0.8723 - 3s/epoch - 25ms/step\n","Epoch 168/300\n","118/118 - 3s - loss: 0.1928 - accuracy: 0.9324 - val_loss: 0.4097 - val_accuracy: 0.8586 - 3s/epoch - 25ms/step\n","Epoch 169/300\n","118/118 - 3s - loss: 0.1907 - accuracy: 0.9332 - val_loss: 0.5157 - val_accuracy: 0.8389 - 3s/epoch - 25ms/step\n","Epoch 170/300\n","118/118 - 3s - loss: 0.1914 - accuracy: 0.9326 - val_loss: 0.4191 - val_accuracy: 0.8582 - 3s/epoch - 25ms/step\n","Epoch 171/300\n","118/118 - 3s - loss: 0.1896 - accuracy: 0.9333 - val_loss: 0.3700 - val_accuracy: 0.8711 - 3s/epoch - 25ms/step\n","Epoch 172/300\n","118/118 - 3s - loss: 0.1887 - accuracy: 0.9341 - val_loss: 0.4337 - val_accuracy: 0.8536 - 3s/epoch - 25ms/step\n","Epoch 173/300\n","118/118 - 3s - loss: 0.1879 - accuracy: 0.9343 - val_loss: 0.4154 - val_accuracy: 0.8585 - 3s/epoch - 25ms/step\n","Epoch 174/300\n","118/118 - 3s - loss: 0.1869 - accuracy: 0.9346 - val_loss: 0.5275 - val_accuracy: 0.8229 - 3s/epoch - 25ms/step\n","Epoch 175/300\n","118/118 - 3s - loss: 0.1874 - accuracy: 0.9347 - val_loss: 0.3837 - val_accuracy: 0.8677 - 3s/epoch - 25ms/step\n","Epoch 176/300\n","118/118 - 3s - loss: 0.1852 - accuracy: 0.9349 - val_loss: 0.3977 - val_accuracy: 0.8661 - 3s/epoch - 25ms/step\n","Epoch 177/300\n","118/118 - 3s - loss: 0.1860 - accuracy: 0.9356 - val_loss: 0.3888 - val_accuracy: 0.8702 - 3s/epoch - 25ms/step\n","Epoch 178/300\n","118/118 - 3s - loss: 0.1838 - accuracy: 0.9355 - val_loss: 0.3915 - val_accuracy: 0.8692 - 3s/epoch - 25ms/step\n","Epoch 179/300\n","118/118 - 3s - loss: 0.1841 - accuracy: 0.9348 - val_loss: 0.4038 - val_accuracy: 0.8638 - 3s/epoch - 25ms/step\n","Epoch 180/300\n","118/118 - 3s - loss: 0.1823 - accuracy: 0.9352 - val_loss: 0.5220 - val_accuracy: 0.8276 - 3s/epoch - 25ms/step\n","Epoch 181/300\n","118/118 - 3s - loss: 0.1826 - accuracy: 0.9367 - val_loss: 0.3819 - val_accuracy: 0.8691 - 3s/epoch - 25ms/step\n","Epoch 182/300\n","118/118 - 3s - loss: 0.1804 - accuracy: 0.9368 - val_loss: 0.4016 - val_accuracy: 0.8638 - 3s/epoch - 25ms/step\n","Epoch 183/300\n","118/118 - 3s - loss: 0.1810 - accuracy: 0.9363 - val_loss: 0.4346 - val_accuracy: 0.8524 - 3s/epoch - 25ms/step\n","Epoch 184/300\n","118/118 - 3s - loss: 0.1807 - accuracy: 0.9360 - val_loss: 0.4230 - val_accuracy: 0.8578 - 3s/epoch - 25ms/step\n","Epoch 185/300\n","118/118 - 3s - loss: 0.1791 - accuracy: 0.9380 - val_loss: 0.3838 - val_accuracy: 0.8721 - 3s/epoch - 25ms/step\n","Epoch 186/300\n","118/118 - 3s - loss: 0.1775 - accuracy: 0.9386 - val_loss: 0.3929 - val_accuracy: 0.8696 - 3s/epoch - 25ms/step\n","Epoch 187/300\n","118/118 - 3s - loss: 0.1773 - accuracy: 0.9383 - val_loss: 0.6375 - val_accuracy: 0.7958 - 3s/epoch - 25ms/step\n","Epoch 188/300\n","118/118 - 3s - loss: 0.1772 - accuracy: 0.9387 - val_loss: 0.3967 - val_accuracy: 0.8683 - 3s/epoch - 25ms/step\n","Epoch 189/300\n","118/118 - 3s - loss: 0.1752 - accuracy: 0.9388 - val_loss: 0.4685 - val_accuracy: 0.8491 - 3s/epoch - 25ms/step\n","Epoch 190/300\n","118/118 - 3s - loss: 0.1736 - accuracy: 0.9395 - val_loss: 0.4233 - val_accuracy: 0.8614 - 3s/epoch - 25ms/step\n","Epoch 191/300\n","118/118 - 3s - loss: 0.1745 - accuracy: 0.9390 - val_loss: 0.4129 - val_accuracy: 0.8628 - 3s/epoch - 25ms/step\n","Epoch 192/300\n","118/118 - 3s - loss: 0.1730 - accuracy: 0.9398 - val_loss: 0.4042 - val_accuracy: 0.8638 - 3s/epoch - 25ms/step\n","Epoch 193/300\n","118/118 - 3s - loss: 0.1718 - accuracy: 0.9403 - val_loss: 0.4207 - val_accuracy: 0.8636 - 3s/epoch - 25ms/step\n","Epoch 194/300\n","118/118 - 3s - loss: 0.1709 - accuracy: 0.9402 - val_loss: 0.4648 - val_accuracy: 0.8411 - 3s/epoch - 25ms/step\n","Epoch 195/300\n","118/118 - 3s - loss: 0.1716 - accuracy: 0.9395 - val_loss: 0.3766 - val_accuracy: 0.8752 - 3s/epoch - 27ms/step\n","Epoch 196/300\n","118/118 - 3s - loss: 0.1712 - accuracy: 0.9405 - val_loss: 0.3773 - val_accuracy: 0.8742 - 3s/epoch - 29ms/step\n","Epoch 197/300\n","118/118 - 3s - loss: 0.1696 - accuracy: 0.9406 - val_loss: 0.3972 - val_accuracy: 0.8669 - 3s/epoch - 25ms/step\n","Epoch 198/300\n","118/118 - 3s - loss: 0.1685 - accuracy: 0.9414 - val_loss: 0.3831 - val_accuracy: 0.8738 - 3s/epoch - 25ms/step\n","Epoch 199/300\n","118/118 - 3s - loss: 0.1669 - accuracy: 0.9417 - val_loss: 0.4292 - val_accuracy: 0.8574 - 3s/epoch - 25ms/step\n","Epoch 200/300\n","118/118 - 3s - loss: 0.1674 - accuracy: 0.9415 - val_loss: 0.5533 - val_accuracy: 0.8270 - 3s/epoch - 25ms/step\n","Epoch 201/300\n","118/118 - 3s - loss: 0.1677 - accuracy: 0.9412 - val_loss: 0.5603 - val_accuracy: 0.8261 - 3s/epoch - 25ms/step\n","Epoch 202/300\n","118/118 - 3s - loss: 0.1681 - accuracy: 0.9415 - val_loss: 0.3939 - val_accuracy: 0.8699 - 3s/epoch - 25ms/step\n","Epoch 203/300\n","118/118 - 3s - loss: 0.1651 - accuracy: 0.9424 - val_loss: 0.4048 - val_accuracy: 0.8663 - 3s/epoch - 25ms/step\n","Epoch 204/300\n","118/118 - 3s - loss: 0.1640 - accuracy: 0.9424 - val_loss: 0.3847 - val_accuracy: 0.8757 - 3s/epoch - 25ms/step\n","Epoch 205/300\n","118/118 - 3s - loss: 0.1625 - accuracy: 0.9435 - val_loss: 0.6587 - val_accuracy: 0.7993 - 3s/epoch - 25ms/step\n","Epoch 206/300\n","118/118 - 3s - loss: 0.1654 - accuracy: 0.9426 - val_loss: 0.4679 - val_accuracy: 0.8493 - 3s/epoch - 25ms/step\n","Epoch 207/300\n","118/118 - 3s - loss: 0.1615 - accuracy: 0.9436 - val_loss: 0.4119 - val_accuracy: 0.8651 - 3s/epoch - 25ms/step\n","Epoch 208/300\n","118/118 - 3s - loss: 0.1619 - accuracy: 0.9431 - val_loss: 0.3973 - val_accuracy: 0.8697 - 3s/epoch - 25ms/step\n","Epoch 209/300\n","118/118 - 3s - loss: 0.1608 - accuracy: 0.9441 - val_loss: 0.4183 - val_accuracy: 0.8608 - 3s/epoch - 25ms/step\n","Epoch 210/300\n","118/118 - 3s - loss: 0.1604 - accuracy: 0.9447 - val_loss: 0.3833 - val_accuracy: 0.8754 - 3s/epoch - 24ms/step\n","Epoch 211/300\n","118/118 - 3s - loss: 0.1599 - accuracy: 0.9449 - val_loss: 0.4004 - val_accuracy: 0.8702 - 3s/epoch - 25ms/step\n","Epoch 212/300\n","118/118 - 3s - loss: 0.1589 - accuracy: 0.9442 - val_loss: 0.3924 - val_accuracy: 0.8709 - 3s/epoch - 25ms/step\n","Epoch 213/300\n","118/118 - 3s - loss: 0.1571 - accuracy: 0.9450 - val_loss: 0.3880 - val_accuracy: 0.8725 - 3s/epoch - 25ms/step\n","Epoch 214/300\n","118/118 - 3s - loss: 0.1556 - accuracy: 0.9454 - val_loss: 0.4307 - val_accuracy: 0.8601 - 3s/epoch - 25ms/step\n","Epoch 215/300\n","118/118 - 3s - loss: 0.1574 - accuracy: 0.9446 - val_loss: 0.4072 - val_accuracy: 0.8683 - 3s/epoch - 25ms/step\n","Epoch 216/300\n","118/118 - 3s - loss: 0.1564 - accuracy: 0.9457 - val_loss: 0.4259 - val_accuracy: 0.8644 - 3s/epoch - 25ms/step\n","Epoch 217/300\n","118/118 - 3s - loss: 0.1562 - accuracy: 0.9446 - val_loss: 0.4367 - val_accuracy: 0.8559 - 3s/epoch - 25ms/step\n","Epoch 218/300\n","118/118 - 3s - loss: 0.1545 - accuracy: 0.9461 - val_loss: 0.4366 - val_accuracy: 0.8596 - 3s/epoch - 25ms/step\n","Epoch 219/300\n","118/118 - 3s - loss: 0.1540 - accuracy: 0.9465 - val_loss: 0.4353 - val_accuracy: 0.8590 - 3s/epoch - 25ms/step\n","Epoch 220/300\n","118/118 - 3s - loss: 0.1544 - accuracy: 0.9459 - val_loss: 0.4725 - val_accuracy: 0.8471 - 3s/epoch - 25ms/step\n","Epoch 221/300\n","118/118 - 3s - loss: 0.1535 - accuracy: 0.9467 - val_loss: 0.4174 - val_accuracy: 0.8689 - 3s/epoch - 25ms/step\n","Epoch 222/300\n","118/118 - 3s - loss: 0.1516 - accuracy: 0.9480 - val_loss: 0.4401 - val_accuracy: 0.8580 - 3s/epoch - 25ms/step\n","Epoch 223/300\n","118/118 - 3s - loss: 0.1506 - accuracy: 0.9477 - val_loss: 0.4165 - val_accuracy: 0.8635 - 3s/epoch - 25ms/step\n","Epoch 224/300\n","118/118 - 3s - loss: 0.1497 - accuracy: 0.9476 - val_loss: 0.4090 - val_accuracy: 0.8688 - 3s/epoch - 25ms/step\n","Epoch 225/300\n","118/118 - 3s - loss: 0.1487 - accuracy: 0.9489 - val_loss: 0.4085 - val_accuracy: 0.8676 - 3s/epoch - 25ms/step\n","Epoch 226/300\n","118/118 - 3s - loss: 0.1484 - accuracy: 0.9491 - val_loss: 0.3887 - val_accuracy: 0.8764 - 3s/epoch - 25ms/step\n","Epoch 227/300\n","118/118 - 3s - loss: 0.1486 - accuracy: 0.9483 - val_loss: 0.4574 - val_accuracy: 0.8552 - 3s/epoch - 25ms/step\n","Epoch 228/300\n","118/118 - 3s - loss: 0.1475 - accuracy: 0.9487 - val_loss: 0.8839 - val_accuracy: 0.7676 - 3s/epoch - 25ms/step\n","Epoch 229/300\n","118/118 - 3s - loss: 0.1506 - accuracy: 0.9480 - val_loss: 0.4064 - val_accuracy: 0.8706 - 3s/epoch - 25ms/step\n","Epoch 230/300\n","118/118 - 3s - loss: 0.1459 - accuracy: 0.9503 - val_loss: 0.4104 - val_accuracy: 0.8664 - 3s/epoch - 25ms/step\n","Epoch 231/300\n","118/118 - 3s - loss: 0.1449 - accuracy: 0.9494 - val_loss: 0.3944 - val_accuracy: 0.8755 - 3s/epoch - 25ms/step\n","Epoch 232/300\n","118/118 - 3s - loss: 0.1444 - accuracy: 0.9503 - val_loss: 0.4302 - val_accuracy: 0.8639 - 3s/epoch - 25ms/step\n","Epoch 233/300\n","118/118 - 3s - loss: 0.1441 - accuracy: 0.9508 - val_loss: 0.3964 - val_accuracy: 0.8725 - 3s/epoch - 25ms/step\n","Epoch 234/300\n","118/118 - 3s - loss: 0.1417 - accuracy: 0.9517 - val_loss: 0.4638 - val_accuracy: 0.8506 - 3s/epoch - 25ms/step\n","Epoch 235/300\n","118/118 - 3s - loss: 0.1434 - accuracy: 0.9505 - val_loss: 0.4595 - val_accuracy: 0.8515 - 3s/epoch - 25ms/step\n","Epoch 236/300\n","118/118 - 3s - loss: 0.1425 - accuracy: 0.9515 - val_loss: 0.4227 - val_accuracy: 0.8679 - 3s/epoch - 25ms/step\n","Epoch 237/300\n","118/118 - 3s - loss: 0.1406 - accuracy: 0.9516 - val_loss: 0.4327 - val_accuracy: 0.8657 - 3s/epoch - 25ms/step\n","Epoch 238/300\n","118/118 - 3s - loss: 0.1399 - accuracy: 0.9516 - val_loss: 0.4773 - val_accuracy: 0.8475 - 3s/epoch - 25ms/step\n","Epoch 239/300\n","118/118 - 3s - loss: 0.1398 - accuracy: 0.9513 - val_loss: 0.5526 - val_accuracy: 0.8396 - 3s/epoch - 25ms/step\n","Epoch 240/300\n","118/118 - 3s - loss: 0.1415 - accuracy: 0.9506 - val_loss: 0.4327 - val_accuracy: 0.8664 - 3s/epoch - 25ms/step\n","Epoch 241/300\n","118/118 - 3s - loss: 0.1388 - accuracy: 0.9524 - val_loss: 0.4805 - val_accuracy: 0.8524 - 3s/epoch - 25ms/step\n","Epoch 242/300\n","118/118 - 3s - loss: 0.1392 - accuracy: 0.9522 - val_loss: 0.4921 - val_accuracy: 0.8482 - 3s/epoch - 25ms/step\n","Epoch 243/300\n","118/118 - 3s - loss: 0.1383 - accuracy: 0.9526 - val_loss: 0.4985 - val_accuracy: 0.8431 - 3s/epoch - 25ms/step\n","Epoch 244/300\n","118/118 - 3s - loss: 0.1379 - accuracy: 0.9525 - val_loss: 0.4218 - val_accuracy: 0.8690 - 3s/epoch - 25ms/step\n","Epoch 245/300\n","118/118 - 3s - loss: 0.1355 - accuracy: 0.9537 - val_loss: 0.4655 - val_accuracy: 0.8561 - 3s/epoch - 25ms/step\n","Epoch 246/300\n","118/118 - 3s - loss: 0.1362 - accuracy: 0.9532 - val_loss: 0.4206 - val_accuracy: 0.8699 - 3s/epoch - 25ms/step\n","Epoch 247/300\n","118/118 - 3s - loss: 0.1354 - accuracy: 0.9529 - val_loss: 0.4617 - val_accuracy: 0.8562 - 3s/epoch - 25ms/step\n","Epoch 248/300\n","118/118 - 3s - loss: 0.1352 - accuracy: 0.9534 - val_loss: 0.4128 - val_accuracy: 0.8721 - 3s/epoch - 25ms/step\n","Epoch 249/300\n","118/118 - 3s - loss: 0.1339 - accuracy: 0.9543 - val_loss: 0.5738 - val_accuracy: 0.8319 - 3s/epoch - 25ms/step\n","Epoch 250/300\n","118/118 - 3s - loss: 0.1354 - accuracy: 0.9529 - val_loss: 0.4366 - val_accuracy: 0.8651 - 3s/epoch - 25ms/step\n","Epoch 251/300\n","118/118 - 3s - loss: 0.1320 - accuracy: 0.9546 - val_loss: 0.4429 - val_accuracy: 0.8657 - 3s/epoch - 25ms/step\n","Epoch 252/300\n","118/118 - 3s - loss: 0.1333 - accuracy: 0.9546 - val_loss: 0.4144 - val_accuracy: 0.8719 - 3s/epoch - 25ms/step\n","Epoch 253/300\n","118/118 - 3s - loss: 0.1315 - accuracy: 0.9552 - val_loss: 0.4333 - val_accuracy: 0.8697 - 3s/epoch - 25ms/step\n","Epoch 254/300\n","118/118 - 3s - loss: 0.1319 - accuracy: 0.9542 - val_loss: 0.4117 - val_accuracy: 0.8735 - 3s/epoch - 25ms/step\n","Epoch 255/300\n","118/118 - 3s - loss: 0.1299 - accuracy: 0.9549 - val_loss: 0.4415 - val_accuracy: 0.8629 - 3s/epoch - 25ms/step\n","Epoch 256/300\n","118/118 - 3s - loss: 0.1285 - accuracy: 0.9566 - val_loss: 0.4246 - val_accuracy: 0.8660 - 3s/epoch - 25ms/step\n","Epoch 257/300\n","118/118 - 3s - loss: 0.1293 - accuracy: 0.9556 - val_loss: 0.4449 - val_accuracy: 0.8693 - 3s/epoch - 25ms/step\n","Epoch 258/300\n","118/118 - 3s - loss: 0.1288 - accuracy: 0.9566 - val_loss: 0.4812 - val_accuracy: 0.8518 - 3s/epoch - 25ms/step\n","Epoch 259/300\n","118/118 - 3s - loss: 0.1290 - accuracy: 0.9563 - val_loss: 0.4874 - val_accuracy: 0.8499 - 3s/epoch - 25ms/step\n","Epoch 260/300\n","118/118 - 3s - loss: 0.1285 - accuracy: 0.9562 - val_loss: 0.4921 - val_accuracy: 0.8446 - 3s/epoch - 25ms/step\n","Epoch 261/300\n","118/118 - 3s - loss: 0.1266 - accuracy: 0.9576 - val_loss: 0.4519 - val_accuracy: 0.8625 - 3s/epoch - 25ms/step\n","Epoch 262/300\n","118/118 - 3s - loss: 0.1263 - accuracy: 0.9572 - val_loss: 0.4204 - val_accuracy: 0.8709 - 3s/epoch - 25ms/step\n","Epoch 263/300\n","118/118 - 3s - loss: 0.1237 - accuracy: 0.9577 - val_loss: 0.4682 - val_accuracy: 0.8596 - 3s/epoch - 26ms/step\n","Epoch 264/300\n","118/118 - 3s - loss: 0.1252 - accuracy: 0.9579 - val_loss: 0.5835 - val_accuracy: 0.8340 - 3s/epoch - 25ms/step\n","Epoch 265/300\n","118/118 - 3s - loss: 0.1232 - accuracy: 0.9586 - val_loss: 0.4558 - val_accuracy: 0.8616 - 3s/epoch - 25ms/step\n","Epoch 266/300\n","118/118 - 3s - loss: 0.1233 - accuracy: 0.9576 - val_loss: 0.5170 - val_accuracy: 0.8501 - 3s/epoch - 25ms/step\n","Epoch 267/300\n","118/118 - 3s - loss: 0.1233 - accuracy: 0.9577 - val_loss: 0.4254 - val_accuracy: 0.8687 - 3s/epoch - 25ms/step\n","Epoch 268/300\n","118/118 - 3s - loss: 0.1233 - accuracy: 0.9575 - val_loss: 0.4336 - val_accuracy: 0.8659 - 3s/epoch - 25ms/step\n","Epoch 269/300\n","118/118 - 3s - loss: 0.1209 - accuracy: 0.9589 - val_loss: 1.1048 - val_accuracy: 0.7642 - 3s/epoch - 25ms/step\n","Epoch 270/300\n","118/118 - 3s - loss: 0.1236 - accuracy: 0.9586 - val_loss: 0.4532 - val_accuracy: 0.8642 - 3s/epoch - 25ms/step\n","Epoch 271/300\n","118/118 - 3s - loss: 0.1214 - accuracy: 0.9591 - val_loss: 0.4812 - val_accuracy: 0.8590 - 3s/epoch - 25ms/step\n","Epoch 272/300\n","118/118 - 4s - loss: 0.1206 - accuracy: 0.9592 - val_loss: 0.4372 - val_accuracy: 0.8640 - 4s/epoch - 30ms/step\n","Epoch 273/300\n","118/118 - 3s - loss: 0.1189 - accuracy: 0.9606 - val_loss: 0.4914 - val_accuracy: 0.8545 - 3s/epoch - 25ms/step\n","Epoch 274/300\n","118/118 - 3s - loss: 0.1193 - accuracy: 0.9599 - val_loss: 0.5019 - val_accuracy: 0.8470 - 3s/epoch - 25ms/step\n","Epoch 275/300\n","118/118 - 3s - loss: 0.1185 - accuracy: 0.9606 - val_loss: 0.5140 - val_accuracy: 0.8543 - 3s/epoch - 25ms/step\n","Epoch 276/300\n","118/118 - 3s - loss: 0.1194 - accuracy: 0.9595 - val_loss: 0.4235 - val_accuracy: 0.8709 - 3s/epoch - 25ms/step\n","Epoch 277/300\n","118/118 - 3s - loss: 0.1196 - accuracy: 0.9595 - val_loss: 0.5554 - val_accuracy: 0.8310 - 3s/epoch - 25ms/step\n","Epoch 278/300\n","118/118 - 3s - loss: 0.1173 - accuracy: 0.9607 - val_loss: 0.5595 - val_accuracy: 0.8360 - 3s/epoch - 25ms/step\n","Epoch 279/300\n","118/118 - 3s - loss: 0.1162 - accuracy: 0.9615 - val_loss: 0.5413 - val_accuracy: 0.8379 - 3s/epoch - 25ms/step\n","Epoch 280/300\n","118/118 - 3s - loss: 0.1159 - accuracy: 0.9601 - val_loss: 0.4504 - val_accuracy: 0.8660 - 3s/epoch - 25ms/step\n","Epoch 281/300\n","118/118 - 3s - loss: 0.1144 - accuracy: 0.9620 - val_loss: 0.5752 - val_accuracy: 0.8297 - 3s/epoch - 25ms/step\n","Epoch 282/300\n","118/118 - 3s - loss: 0.1150 - accuracy: 0.9611 - val_loss: 0.4309 - val_accuracy: 0.8715 - 3s/epoch - 25ms/step\n","Epoch 283/300\n","118/118 - 3s - loss: 0.1171 - accuracy: 0.9599 - val_loss: 0.7396 - val_accuracy: 0.7968 - 3s/epoch - 25ms/step\n","Epoch 284/300\n","118/118 - 3s - loss: 0.1154 - accuracy: 0.9613 - val_loss: 0.4618 - val_accuracy: 0.8611 - 3s/epoch - 25ms/step\n","Epoch 285/300\n","118/118 - 3s - loss: 0.1138 - accuracy: 0.9620 - val_loss: 0.4266 - val_accuracy: 0.8766 - 3s/epoch - 25ms/step\n","Epoch 286/300\n","118/118 - 3s - loss: 0.1122 - accuracy: 0.9626 - val_loss: 0.4903 - val_accuracy: 0.8578 - 3s/epoch - 25ms/step\n","Epoch 287/300\n","118/118 - 3s - loss: 0.1116 - accuracy: 0.9627 - val_loss: 0.4505 - val_accuracy: 0.8686 - 3s/epoch - 25ms/step\n","Epoch 288/300\n","118/118 - 3s - loss: 0.1118 - accuracy: 0.9628 - val_loss: 0.4697 - val_accuracy: 0.8595 - 3s/epoch - 25ms/step\n","Epoch 289/300\n","118/118 - 3s - loss: 0.1120 - accuracy: 0.9624 - val_loss: 0.4729 - val_accuracy: 0.8654 - 3s/epoch - 25ms/step\n","Epoch 290/300\n","118/118 - 3s - loss: 0.1114 - accuracy: 0.9630 - val_loss: 0.4235 - val_accuracy: 0.8704 - 3s/epoch - 25ms/step\n","Epoch 291/300\n","118/118 - 3s - loss: 0.1104 - accuracy: 0.9627 - val_loss: 0.4775 - val_accuracy: 0.8691 - 3s/epoch - 25ms/step\n","Epoch 292/300\n","118/118 - 3s - loss: 0.1087 - accuracy: 0.9637 - val_loss: 0.6491 - val_accuracy: 0.8256 - 3s/epoch - 25ms/step\n","Epoch 293/300\n","118/118 - 3s - loss: 0.1095 - accuracy: 0.9640 - val_loss: 0.4378 - val_accuracy: 0.8693 - 3s/epoch - 25ms/step\n","Epoch 294/300\n","118/118 - 3s - loss: 0.1074 - accuracy: 0.9639 - val_loss: 0.5656 - val_accuracy: 0.8406 - 3s/epoch - 25ms/step\n","Epoch 295/300\n","118/118 - 3s - loss: 0.1075 - accuracy: 0.9641 - val_loss: 0.4650 - val_accuracy: 0.8656 - 3s/epoch - 25ms/step\n","Epoch 296/300\n","118/118 - 3s - loss: 0.1081 - accuracy: 0.9643 - val_loss: 0.5102 - val_accuracy: 0.8572 - 3s/epoch - 26ms/step\n","Epoch 297/300\n","118/118 - 3s - loss: 0.1099 - accuracy: 0.9627 - val_loss: 0.4284 - val_accuracy: 0.8754 - 3s/epoch - 26ms/step\n","Epoch 298/300\n","118/118 - 3s - loss: 0.1040 - accuracy: 0.9656 - val_loss: 0.4353 - val_accuracy: 0.8714 - 3s/epoch - 25ms/step\n","Epoch 299/300\n","118/118 - 3s - loss: 0.1046 - accuracy: 0.9654 - val_loss: 0.4420 - val_accuracy: 0.8688 - 3s/epoch - 25ms/step\n","Epoch 300/300\n","118/118 - 3s - loss: 0.1052 - accuracy: 0.9655 - val_loss: 0.5124 - val_accuracy: 0.8534 - 3s/epoch - 26ms/step\n"]}],"source":["# solution 4: Normalization inside network\n","\n","initializer = tf.keras.initializers.RandomNormal()\n","\n","#################### YOUR CODE HERE ####################\n","# Các bạn tìm cách code để add BatchNormalization layer\n","# trước mỗi hidden layer\n","model = keras.Sequential()\n","model.add(keras.Input(shape=(784,)))\n","model.add(keras.layers.BatchNormalization())\n","model.add(keras.layers.Dense(128, activation='sigmoid',\n","                             kernel_initializer=initializer))\n","model.add(keras.layers.BatchNormalization())\n","model.add(keras.layers.Dense(128, activation='sigmoid',\n","                             kernel_initializer=initializer))\n","model.add(keras.layers.BatchNormalization())\n","model.add(keras.layers.Dense(128, activation='sigmoid',\n","                             kernel_initializer=initializer))\n","model.add(keras.layers.BatchNormalization())\n","model.add(keras.layers.Dense(128, activation='sigmoid',\n","                             kernel_initializer=initializer))\n","model.add(keras.layers.BatchNormalization())\n","model.add(keras.layers.Dense(128, activation='sigmoid',\n","                             kernel_initializer=initializer))\n","model.add(keras.layers.BatchNormalization())\n","model.add(keras.layers.Dense(128, activation='sigmoid',\n","                             kernel_initializer=initializer))\n","model.add(keras.layers.BatchNormalization())\n","model.add(keras.layers.Dense(128, activation='sigmoid',\n","                             kernel_initializer=initializer))\n","model.add(keras.layers.Dense(10))\n","########################################################\n","\n","# optimizer and loss\n","model.compile(optimizer=tf.keras.optimizers.SGD(), \n","              loss=keras.losses.SparseCategoricalCrossentropy(from_logits=True), \n","              metrics=['accuracy'])\n","\n","# training\n","batch_size = 512\n","history = model.fit(X_train, y_train, batch_size,\n","                    validation_data=(X_test, y_test),\n","                    epochs=300, verbose=2)"]},{"cell_type":"code","execution_count":11,"id":"4da831e7","metadata":{"id":"4da831e7","executionInfo":{"status":"ok","timestamp":1669897657981,"user_tz":-420,"elapsed":28,"user":{"displayName":"Thanh Hùng","userId":"08335145449228425002"}}},"outputs":[],"source":[]},{"cell_type":"code","execution_count":12,"id":"248dd07e","metadata":{"colab":{"base_uri":"https://localhost:8080/","height":296},"id":"248dd07e","executionInfo":{"status":"ok","timestamp":1669897657982,"user_tz":-420,"elapsed":19,"user":{"displayName":"Thanh Hùng","userId":"08335145449228425002"}},"outputId":"4564ea2f-2345-43c0-bdb2-0d66060a7908"},"outputs":[{"output_type":"execute_result","data":{"text/plain":["<matplotlib.legend.Legend at 0x7fb556d03c70>"]},"metadata":{},"execution_count":12},{"output_type":"display_data","data":{"text/plain":["<Figure size 432x288 with 1 Axes>"],"image/png":"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\n"},"metadata":{"needs_background":"light"}}],"source":["import matplotlib.pyplot as plt\n","\n","plt.plot(history.history['loss'], label='loss')\n","plt.plot(history.history['val_loss'], label='val_loss')\n","plt.xlabel('iteration')\n","plt.ylabel('Loss')\n","plt.legend()"]},{"cell_type":"code","execution_count":12,"id":"8e35d7eb","metadata":{"id":"8e35d7eb","executionInfo":{"status":"ok","timestamp":1669897657982,"user_tz":-420,"elapsed":8,"user":{"displayName":"Thanh Hùng","userId":"08335145449228425002"}}},"outputs":[],"source":[]},{"cell_type":"code","execution_count":13,"id":"4b997d5b","metadata":{"colab":{"base_uri":"https://localhost:8080/","height":298},"id":"4b997d5b","executionInfo":{"status":"ok","timestamp":1669897658785,"user_tz":-420,"elapsed":811,"user":{"displayName":"Thanh Hùng","userId":"08335145449228425002"}},"outputId":"77aa80d8-e7ee-4bf0-c969-6504dbaf0871"},"outputs":[{"output_type":"execute_result","data":{"text/plain":["<matplotlib.legend.Legend at 0x7fb5575e2af0>"]},"metadata":{},"execution_count":13},{"output_type":"display_data","data":{"text/plain":["<Figure size 432x288 with 1 Axes>"],"image/png":"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\n"},"metadata":{"needs_background":"light"}}],"source":["import matplotlib.pyplot as plt\n","\n","plt.plot(history.history['accuracy'], label='train_accuracy')\n","plt.plot(history.history['val_accuracy'], label='val_accuracy')\n","plt.xlabel('iteration')\n","plt.ylabel('Accuracy')\n","plt.legend()"]},{"cell_type":"code","execution_count":13,"id":"8623181e","metadata":{"id":"8623181e","executionInfo":{"status":"ok","timestamp":1669897658786,"user_tz":-420,"elapsed":8,"user":{"displayName":"Thanh Hùng","userId":"08335145449228425002"}}},"outputs":[],"source":[]},{"cell_type":"code","execution_count":13,"id":"c84dafda","metadata":{"id":"c84dafda","executionInfo":{"status":"ok","timestamp":1669897658786,"user_tz":-420,"elapsed":7,"user":{"displayName":"Thanh Hùng","userId":"08335145449228425002"}}},"outputs":[],"source":[]},{"cell_type":"code","execution_count":13,"id":"c3b045b0","metadata":{"id":"c3b045b0","executionInfo":{"status":"ok","timestamp":1669897658787,"user_tz":-420,"elapsed":8,"user":{"displayName":"Thanh Hùng","userId":"08335145449228425002"}}},"outputs":[],"source":[]}],"metadata":{"kernelspec":{"display_name":"Python 3","language":"python","name":"python3"},"language_info":{"codemirror_mode":{"name":"ipython","version":3},"file_extension":".py","mimetype":"text/x-python","name":"python","nbconvert_exporter":"python","pygments_lexer":"ipython3","version":"3.8.8"},"colab":{"provenance":[],"toc_visible":true},"gpuClass":"standard"},"nbformat":4,"nbformat_minor":5}