[hint]Solution2_BetteActivationFunction.ipynb

{"cells":[{"cell_type":"markdown","source":["## Load Data"],"metadata":{"id":"7JzDbACfT3rI"},"id":"7JzDbACfT3rI"},{"cell_type":"code","execution_count":1,"id":"e175eeac","metadata":{"colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"elapsed":4100,"status":"ok","timestamp":1669890036996,"user":{"displayName":"Thanh Hùng","userId":"08335145449228425002"},"user_tz":-420},"id":"e175eeac","outputId":"6478f279-2547-43b5-b9d7-63f6fd141a71"},"outputs":[{"output_type":"stream","name":"stdout","text":["Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/train-labels-idx1-ubyte.gz\n","29515/29515 [==============================] - 0s 0us/step\n","Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/train-images-idx3-ubyte.gz\n","26421880/26421880 [==============================] - 0s 0us/step\n","Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/t10k-labels-idx1-ubyte.gz\n","5148/5148 [==============================] - 0s 0us/step\n","Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/t10k-images-idx3-ubyte.gz\n","4422102/4422102 [==============================] - 0s 0us/step\n","(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":2,"id":"8aabf29a","metadata":{"colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"elapsed":6,"status":"ok","timestamp":1669890036997,"user":{"displayName":"Thanh Hùng","userId":"08335145449228425002"},"user_tz":-420},"id":"8aabf29a","outputId":"c2405dba-0e83-4637-ff39-0074c26f2970"},"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":["## Better Activation"],"metadata":{"id":"goM2LmCDVEzh"},"id":"goM2LmCDVEzh"},{"cell_type":"code","execution_count":3,"id":"VnOXPEfY6XtS","metadata":{"id":"VnOXPEfY6XtS","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1669890360940,"user_tz":-420,"elapsed":323946,"user":{"displayName":"Thanh Hùng","userId":"08335145449228425002"}},"outputId":"dc29368e-080c-4870-b6df-a8098a35bab4"},"outputs":[{"output_type":"stream","name":"stdout","text":["Epoch 1/150\n","118/118 - 5s - loss: 1.1096 - accuracy: 0.6076 - val_loss: 0.8095 - val_accuracy: 0.7060 - 5s/epoch - 46ms/step\n","Epoch 2/150\n","118/118 - 3s - loss: 0.6440 - accuracy: 0.7640 - val_loss: 0.6419 - val_accuracy: 0.7575 - 3s/epoch - 23ms/step\n","Epoch 3/150\n","118/118 - 2s - loss: 0.5590 - accuracy: 0.7967 - val_loss: 0.7203 - val_accuracy: 0.7186 - 2s/epoch - 18ms/step\n","Epoch 4/150\n","118/118 - 2s - loss: 0.5131 - accuracy: 0.8152 - val_loss: 0.5644 - val_accuracy: 0.7924 - 2s/epoch - 17ms/step\n","Epoch 5/150\n","118/118 - 2s - loss: 0.4787 - accuracy: 0.8271 - val_loss: 0.4699 - val_accuracy: 0.8329 - 2s/epoch - 17ms/step\n","Epoch 6/150\n","118/118 - 2s - loss: 0.4603 - accuracy: 0.8333 - val_loss: 0.6138 - val_accuracy: 0.7858 - 2s/epoch - 17ms/step\n","Epoch 7/150\n","118/118 - 2s - loss: 0.4431 - accuracy: 0.8413 - val_loss: 0.5068 - val_accuracy: 0.8012 - 2s/epoch - 17ms/step\n","Epoch 8/150\n","118/118 - 2s - loss: 0.4268 - accuracy: 0.8443 - val_loss: 0.5998 - val_accuracy: 0.7825 - 2s/epoch - 17ms/step\n","Epoch 9/150\n","118/118 - 2s - loss: 0.4087 - accuracy: 0.8515 - val_loss: 0.7439 - val_accuracy: 0.7457 - 2s/epoch - 17ms/step\n","Epoch 10/150\n","118/118 - 2s - loss: 0.4175 - accuracy: 0.8499 - val_loss: 0.5144 - val_accuracy: 0.8134 - 2s/epoch - 17ms/step\n","Epoch 11/150\n","118/118 - 2s - loss: 0.3882 - accuracy: 0.8593 - val_loss: 0.5654 - val_accuracy: 0.7869 - 2s/epoch - 18ms/step\n","Epoch 12/150\n","118/118 - 2s - loss: 0.3811 - accuracy: 0.8609 - val_loss: 0.4858 - val_accuracy: 0.8193 - 2s/epoch - 17ms/step\n","Epoch 13/150\n","118/118 - 2s - loss: 0.3753 - accuracy: 0.8636 - val_loss: 0.4201 - val_accuracy: 0.8492 - 2s/epoch - 17ms/step\n","Epoch 14/150\n","118/118 - 2s - loss: 0.3637 - accuracy: 0.8676 - val_loss: 0.4709 - val_accuracy: 0.8318 - 2s/epoch - 17ms/step\n","Epoch 15/150\n","118/118 - 2s - loss: 0.3632 - accuracy: 0.8674 - val_loss: 0.4635 - val_accuracy: 0.8367 - 2s/epoch - 17ms/step\n","Epoch 16/150\n","118/118 - 2s - loss: 0.3526 - accuracy: 0.8717 - val_loss: 0.4999 - val_accuracy: 0.8245 - 2s/epoch - 17ms/step\n","Epoch 17/150\n","118/118 - 2s - loss: 0.3455 - accuracy: 0.8719 - val_loss: 0.4488 - val_accuracy: 0.8282 - 2s/epoch - 17ms/step\n","Epoch 18/150\n","118/118 - 2s - loss: 0.3421 - accuracy: 0.8759 - val_loss: 0.4151 - val_accuracy: 0.8499 - 2s/epoch - 17ms/step\n","Epoch 19/150\n","118/118 - 2s - loss: 0.3373 - accuracy: 0.8761 - val_loss: 0.4279 - val_accuracy: 0.8418 - 2s/epoch - 17ms/step\n","Epoch 20/150\n","118/118 - 2s - loss: 0.3329 - accuracy: 0.8782 - val_loss: 0.4664 - val_accuracy: 0.8241 - 2s/epoch - 17ms/step\n","Epoch 21/150\n","118/118 - 2s - loss: 0.3299 - accuracy: 0.8793 - val_loss: 0.4321 - val_accuracy: 0.8328 - 2s/epoch - 17ms/step\n","Epoch 22/150\n","118/118 - 2s - loss: 0.3232 - accuracy: 0.8807 - val_loss: 0.5906 - val_accuracy: 0.7994 - 2s/epoch - 17ms/step\n","Epoch 23/150\n","118/118 - 2s - loss: 0.3237 - accuracy: 0.8823 - val_loss: 0.5034 - val_accuracy: 0.8142 - 2s/epoch - 17ms/step\n","Epoch 24/150\n","118/118 - 2s - loss: 0.3166 - accuracy: 0.8831 - val_loss: 0.3952 - val_accuracy: 0.8539 - 2s/epoch - 17ms/step\n","Epoch 25/150\n","118/118 - 2s - loss: 0.3106 - accuracy: 0.8860 - val_loss: 0.3991 - val_accuracy: 0.8560 - 2s/epoch - 17ms/step\n","Epoch 26/150\n","118/118 - 2s - loss: 0.3063 - accuracy: 0.8875 - val_loss: 0.4334 - val_accuracy: 0.8517 - 2s/epoch - 17ms/step\n","Epoch 27/150\n","118/118 - 2s - loss: 0.3055 - accuracy: 0.8880 - val_loss: 0.5224 - val_accuracy: 0.8083 - 2s/epoch - 17ms/step\n","Epoch 28/150\n","118/118 - 2s - loss: 0.2990 - accuracy: 0.8900 - val_loss: 0.4896 - val_accuracy: 0.8271 - 2s/epoch - 17ms/step\n","Epoch 29/150\n","118/118 - 2s - loss: 0.2992 - accuracy: 0.8900 - val_loss: 0.4246 - val_accuracy: 0.8407 - 2s/epoch - 17ms/step\n","Epoch 30/150\n","118/118 - 2s - loss: 0.2933 - accuracy: 0.8920 - val_loss: 0.4668 - val_accuracy: 0.8201 - 2s/epoch - 17ms/step\n","Epoch 31/150\n","118/118 - 2s - loss: 0.2905 - accuracy: 0.8925 - val_loss: 0.4062 - val_accuracy: 0.8533 - 2s/epoch - 17ms/step\n","Epoch 32/150\n","118/118 - 2s - loss: 0.2868 - accuracy: 0.8948 - val_loss: 0.4332 - val_accuracy: 0.8411 - 2s/epoch - 17ms/step\n","Epoch 33/150\n","118/118 - 2s - loss: 0.2855 - accuracy: 0.8932 - val_loss: 0.4058 - val_accuracy: 0.8588 - 2s/epoch - 17ms/step\n","Epoch 34/150\n","118/118 - 2s - loss: 0.2798 - accuracy: 0.8981 - val_loss: 0.4282 - val_accuracy: 0.8421 - 2s/epoch - 17ms/step\n","Epoch 35/150\n","118/118 - 2s - loss: 0.2801 - accuracy: 0.8960 - val_loss: 0.4401 - val_accuracy: 0.8369 - 2s/epoch - 17ms/step\n","Epoch 36/150\n","118/118 - 2s - loss: 0.2771 - accuracy: 0.8968 - val_loss: 0.4346 - val_accuracy: 0.8344 - 2s/epoch - 17ms/step\n","Epoch 37/150\n","118/118 - 2s - loss: 0.2760 - accuracy: 0.8969 - val_loss: 0.3748 - val_accuracy: 0.8645 - 2s/epoch - 17ms/step\n","Epoch 38/150\n","118/118 - 3s - loss: 0.2714 - accuracy: 0.8999 - val_loss: 0.3856 - val_accuracy: 0.8591 - 3s/epoch - 23ms/step\n","Epoch 39/150\n","118/118 - 2s - loss: 0.2707 - accuracy: 0.9001 - val_loss: 0.3903 - val_accuracy: 0.8609 - 2s/epoch - 18ms/step\n","Epoch 40/150\n","118/118 - 2s - loss: 0.2652 - accuracy: 0.9015 - val_loss: 0.4103 - val_accuracy: 0.8546 - 2s/epoch - 17ms/step\n","Epoch 41/150\n","118/118 - 2s - loss: 0.2665 - accuracy: 0.9013 - val_loss: 0.3937 - val_accuracy: 0.8574 - 2s/epoch - 17ms/step\n","Epoch 42/150\n","118/118 - 2s - loss: 0.2636 - accuracy: 0.9029 - val_loss: 0.7715 - val_accuracy: 0.7768 - 2s/epoch - 17ms/step\n","Epoch 43/150\n","118/118 - 2s - loss: 0.2764 - accuracy: 0.9010 - val_loss: 0.4401 - val_accuracy: 0.8483 - 2s/epoch - 17ms/step\n","Epoch 44/150\n","118/118 - 2s - loss: 0.2575 - accuracy: 0.9043 - val_loss: 0.4223 - val_accuracy: 0.8515 - 2s/epoch - 17ms/step\n","Epoch 45/150\n","118/118 - 2s - loss: 0.2509 - accuracy: 0.9074 - val_loss: 0.3963 - val_accuracy: 0.8549 - 2s/epoch - 17ms/step\n","Epoch 46/150\n","118/118 - 2s - loss: 0.2505 - accuracy: 0.9077 - val_loss: 0.4152 - val_accuracy: 0.8461 - 2s/epoch - 17ms/step\n","Epoch 47/150\n","118/118 - 2s - loss: 0.2502 - accuracy: 0.9066 - val_loss: 0.4276 - val_accuracy: 0.8498 - 2s/epoch - 17ms/step\n","Epoch 48/150\n","118/118 - 2s - loss: 0.2450 - accuracy: 0.9090 - val_loss: 0.4216 - val_accuracy: 0.8554 - 2s/epoch - 17ms/step\n","Epoch 49/150\n","118/118 - 2s - loss: 0.2423 - accuracy: 0.9108 - val_loss: 0.3852 - val_accuracy: 0.8607 - 2s/epoch - 17ms/step\n","Epoch 50/150\n","118/118 - 2s - loss: 0.2387 - accuracy: 0.9111 - val_loss: 0.3893 - val_accuracy: 0.8639 - 2s/epoch - 17ms/step\n","Epoch 51/150\n","118/118 - 2s - loss: 0.2388 - accuracy: 0.9117 - val_loss: 0.3699 - val_accuracy: 0.8677 - 2s/epoch - 17ms/step\n","Epoch 52/150\n","118/118 - 2s - loss: 0.2375 - accuracy: 0.9118 - val_loss: 0.4960 - val_accuracy: 0.8368 - 2s/epoch - 17ms/step\n","Epoch 53/150\n","118/118 - 2s - loss: 0.2391 - accuracy: 0.9118 - val_loss: 0.4877 - val_accuracy: 0.8397 - 2s/epoch - 17ms/step\n","Epoch 54/150\n","118/118 - 2s - loss: 0.2408 - accuracy: 0.9106 - val_loss: 0.4198 - val_accuracy: 0.8545 - 2s/epoch - 17ms/step\n","Epoch 55/150\n","118/118 - 2s - loss: 0.2326 - accuracy: 0.9141 - val_loss: 0.4447 - val_accuracy: 0.8463 - 2s/epoch - 17ms/step\n","Epoch 56/150\n","118/118 - 2s - loss: 0.2299 - accuracy: 0.9138 - val_loss: 0.4138 - val_accuracy: 0.8608 - 2s/epoch - 17ms/step\n","Epoch 57/150\n","118/118 - 2s - loss: 0.2260 - accuracy: 0.9160 - val_loss: 0.4172 - val_accuracy: 0.8578 - 2s/epoch - 17ms/step\n","Epoch 58/150\n","118/118 - 2s - loss: 0.2261 - accuracy: 0.9161 - val_loss: 0.4202 - val_accuracy: 0.8502 - 2s/epoch - 17ms/step\n","Epoch 59/150\n","118/118 - 2s - loss: 0.2255 - accuracy: 0.9156 - val_loss: 0.4402 - val_accuracy: 0.8516 - 2s/epoch - 17ms/step\n","Epoch 60/150\n","118/118 - 2s - loss: 0.2225 - accuracy: 0.9172 - val_loss: 0.4547 - val_accuracy: 0.8444 - 2s/epoch - 17ms/step\n","Epoch 61/150\n","118/118 - 2s - loss: 0.2212 - accuracy: 0.9179 - val_loss: 0.4058 - val_accuracy: 0.8600 - 2s/epoch - 17ms/step\n","Epoch 62/150\n","118/118 - 2s - loss: 0.2178 - accuracy: 0.9192 - val_loss: 0.4417 - val_accuracy: 0.8543 - 2s/epoch - 18ms/step\n","Epoch 63/150\n","118/118 - 2s - loss: 0.2207 - accuracy: 0.9171 - val_loss: 0.4313 - val_accuracy: 0.8571 - 2s/epoch - 17ms/step\n","Epoch 64/150\n","118/118 - 2s - loss: 0.2304 - accuracy: 0.9165 - val_loss: 0.4647 - val_accuracy: 0.8489 - 2s/epoch - 17ms/step\n","Epoch 65/150\n","118/118 - 2s - loss: 0.2103 - accuracy: 0.9219 - val_loss: 0.4587 - val_accuracy: 0.8515 - 2s/epoch - 17ms/step\n","Epoch 66/150\n","118/118 - 2s - loss: 0.2109 - accuracy: 0.9216 - val_loss: 0.5848 - val_accuracy: 0.8053 - 2s/epoch - 17ms/step\n","Epoch 67/150\n","118/118 - 2s - loss: 0.2129 - accuracy: 0.9213 - val_loss: 0.3801 - val_accuracy: 0.8725 - 2s/epoch - 17ms/step\n","Epoch 68/150\n","118/118 - 2s - loss: 0.2109 - accuracy: 0.9214 - val_loss: 0.4045 - val_accuracy: 0.8632 - 2s/epoch - 17ms/step\n","Epoch 69/150\n","118/118 - 2s - loss: 0.2022 - accuracy: 0.9243 - val_loss: 0.3731 - val_accuracy: 0.8760 - 2s/epoch - 17ms/step\n","Epoch 70/150\n","118/118 - 2s - loss: 0.2033 - accuracy: 0.9244 - val_loss: 0.4557 - val_accuracy: 0.8524 - 2s/epoch - 17ms/step\n","Epoch 71/150\n","118/118 - 2s - loss: 0.2055 - accuracy: 0.9234 - val_loss: 0.4435 - val_accuracy: 0.8573 - 2s/epoch - 17ms/step\n","Epoch 72/150\n","118/118 - 2s - loss: 0.2019 - accuracy: 0.9243 - val_loss: 0.4951 - val_accuracy: 0.8427 - 2s/epoch - 17ms/step\n","Epoch 73/150\n","118/118 - 2s - loss: 0.1985 - accuracy: 0.9271 - val_loss: 0.3902 - val_accuracy: 0.8680 - 2s/epoch - 17ms/step\n","Epoch 74/150\n","118/118 - 2s - loss: 0.1963 - accuracy: 0.9266 - val_loss: 0.4221 - val_accuracy: 0.8620 - 2s/epoch - 17ms/step\n","Epoch 75/150\n","118/118 - 2s - loss: 0.1938 - accuracy: 0.9284 - val_loss: 0.4314 - val_accuracy: 0.8573 - 2s/epoch - 17ms/step\n","Epoch 76/150\n","118/118 - 2s - loss: 0.1986 - accuracy: 0.9256 - val_loss: 0.4111 - val_accuracy: 0.8686 - 2s/epoch - 17ms/step\n","Epoch 77/150\n","118/118 - 2s - loss: 0.1908 - accuracy: 0.9294 - val_loss: 0.4431 - val_accuracy: 0.8636 - 2s/epoch - 17ms/step\n","Epoch 78/150\n","118/118 - 2s - loss: 0.1865 - accuracy: 0.9306 - val_loss: 0.4650 - val_accuracy: 0.8547 - 2s/epoch - 17ms/step\n","Epoch 79/150\n","118/118 - 2s - loss: 0.1909 - accuracy: 0.9290 - val_loss: 0.4772 - val_accuracy: 0.8353 - 2s/epoch - 17ms/step\n","Epoch 80/150\n","118/118 - 2s - loss: 0.1875 - accuracy: 0.9294 - val_loss: 0.4327 - val_accuracy: 0.8517 - 2s/epoch - 17ms/step\n","Epoch 81/150\n","118/118 - 2s - loss: 0.1862 - accuracy: 0.9309 - val_loss: 0.4159 - val_accuracy: 0.8695 - 2s/epoch - 17ms/step\n","Epoch 82/150\n","118/118 - 2s - loss: 0.1856 - accuracy: 0.9301 - val_loss: 0.4348 - val_accuracy: 0.8534 - 2s/epoch - 17ms/step\n","Epoch 83/150\n","118/118 - 2s - loss: 0.1851 - accuracy: 0.9312 - val_loss: 0.4835 - val_accuracy: 0.8381 - 2s/epoch - 17ms/step\n","Epoch 84/150\n","118/118 - 2s - loss: 0.1851 - accuracy: 0.9308 - val_loss: 0.4882 - val_accuracy: 0.8547 - 2s/epoch - 17ms/step\n","Epoch 85/150\n","118/118 - 2s - loss: 0.1809 - accuracy: 0.9320 - val_loss: 0.4558 - val_accuracy: 0.8571 - 2s/epoch - 17ms/step\n","Epoch 86/150\n","118/118 - 2s - loss: 0.1827 - accuracy: 0.9316 - val_loss: 0.3878 - val_accuracy: 0.8727 - 2s/epoch - 17ms/step\n","Epoch 87/150\n","118/118 - 2s - loss: 0.1764 - accuracy: 0.9349 - val_loss: 0.4095 - val_accuracy: 0.8689 - 2s/epoch - 17ms/step\n","Epoch 88/150\n","118/118 - 2s - loss: 0.1740 - accuracy: 0.9353 - val_loss: 0.4133 - val_accuracy: 0.8742 - 2s/epoch - 17ms/step\n","Epoch 89/150\n","118/118 - 2s - loss: 0.1705 - accuracy: 0.9359 - val_loss: 0.4148 - val_accuracy: 0.8721 - 2s/epoch - 17ms/step\n","Epoch 90/150\n","118/118 - 2s - loss: 0.1768 - accuracy: 0.9343 - val_loss: 0.4413 - val_accuracy: 0.8676 - 2s/epoch - 17ms/step\n","Epoch 91/150\n","118/118 - 2s - loss: 0.1739 - accuracy: 0.9353 - val_loss: 0.3940 - val_accuracy: 0.8729 - 2s/epoch - 17ms/step\n","Epoch 92/150\n","118/118 - 2s - loss: 0.1689 - accuracy: 0.9373 - val_loss: 0.6391 - val_accuracy: 0.8316 - 2s/epoch - 17ms/step\n","Epoch 93/150\n","118/118 - 2s - loss: 0.1796 - accuracy: 0.9336 - val_loss: 0.4057 - val_accuracy: 0.8708 - 2s/epoch - 17ms/step\n","Epoch 94/150\n","118/118 - 2s - loss: 0.1638 - accuracy: 0.9392 - val_loss: 0.4555 - val_accuracy: 0.8613 - 2s/epoch - 17ms/step\n","Epoch 95/150\n","118/118 - 2s - loss: 0.1882 - accuracy: 0.9316 - val_loss: 0.3973 - val_accuracy: 0.8780 - 2s/epoch - 17ms/step\n","Epoch 96/150\n","118/118 - 2s - loss: 0.1629 - accuracy: 0.9397 - val_loss: 0.4595 - val_accuracy: 0.8566 - 2s/epoch - 17ms/step\n","Epoch 97/150\n","118/118 - 2s - loss: 0.1674 - accuracy: 0.9378 - val_loss: 0.5036 - val_accuracy: 0.8498 - 2s/epoch - 17ms/step\n","Epoch 98/150\n","118/118 - 2s - loss: 0.1601 - accuracy: 0.9409 - val_loss: 0.5375 - val_accuracy: 0.8415 - 2s/epoch - 17ms/step\n","Epoch 99/150\n","118/118 - 2s - loss: 0.1645 - accuracy: 0.9396 - val_loss: 0.5437 - val_accuracy: 0.8499 - 2s/epoch - 17ms/step\n","Epoch 100/150\n","118/118 - 2s - loss: 0.1571 - accuracy: 0.9421 - val_loss: 0.4144 - val_accuracy: 0.8719 - 2s/epoch - 17ms/step\n","Epoch 101/150\n","118/118 - 2s - loss: 0.1588 - accuracy: 0.9416 - val_loss: 0.4305 - val_accuracy: 0.8741 - 2s/epoch - 17ms/step\n","Epoch 102/150\n","118/118 - 2s - loss: 0.1577 - accuracy: 0.9419 - val_loss: 0.4694 - val_accuracy: 0.8659 - 2s/epoch - 17ms/step\n","Epoch 103/150\n","118/118 - 2s - loss: 0.1569 - accuracy: 0.9428 - val_loss: 0.4772 - val_accuracy: 0.8518 - 2s/epoch - 17ms/step\n","Epoch 104/150\n","118/118 - 2s - loss: 0.1605 - accuracy: 0.9407 - val_loss: 0.4449 - val_accuracy: 0.8715 - 2s/epoch - 17ms/step\n","Epoch 105/150\n","118/118 - 2s - loss: 0.1511 - accuracy: 0.9434 - val_loss: 0.4232 - val_accuracy: 0.8695 - 2s/epoch - 17ms/step\n","Epoch 106/150\n","118/118 - 2s - loss: 0.1536 - accuracy: 0.9422 - val_loss: 0.4415 - val_accuracy: 0.8720 - 2s/epoch - 17ms/step\n","Epoch 107/150\n","118/118 - 2s - loss: 0.1485 - accuracy: 0.9453 - val_loss: 0.4481 - val_accuracy: 0.8757 - 2s/epoch - 17ms/step\n","Epoch 108/150\n","118/118 - 2s - loss: 0.6787 - accuracy: 0.8238 - val_loss: 0.5463 - val_accuracy: 0.8093 - 2s/epoch - 17ms/step\n","Epoch 109/150\n","118/118 - 2s - loss: 0.3585 - accuracy: 0.8699 - val_loss: 0.4166 - val_accuracy: 0.8497 - 2s/epoch - 18ms/step\n","Epoch 110/150\n","118/118 - 2s - loss: 0.2831 - accuracy: 0.8943 - val_loss: 0.4678 - val_accuracy: 0.8407 - 2s/epoch - 17ms/step\n","Epoch 111/150\n","118/118 - 2s - loss: 0.2487 - accuracy: 0.9069 - val_loss: 0.4527 - val_accuracy: 0.8511 - 2s/epoch - 17ms/step\n","Epoch 112/150\n","118/118 - 2s - loss: 0.2255 - accuracy: 0.9153 - val_loss: 0.4306 - val_accuracy: 0.8582 - 2s/epoch - 17ms/step\n","Epoch 113/150\n","118/118 - 2s - loss: 0.2035 - accuracy: 0.9245 - val_loss: 0.4414 - val_accuracy: 0.8594 - 2s/epoch - 17ms/step\n","Epoch 114/150\n","118/118 - 2s - loss: 0.1934 - accuracy: 0.9271 - val_loss: 0.5838 - val_accuracy: 0.8328 - 2s/epoch - 17ms/step\n","Epoch 115/150\n","118/118 - 2s - loss: 0.1875 - accuracy: 0.9308 - val_loss: 0.4439 - val_accuracy: 0.8577 - 2s/epoch - 17ms/step\n","Epoch 116/150\n","118/118 - 2s - loss: 0.1743 - accuracy: 0.9344 - val_loss: 0.5453 - val_accuracy: 0.8337 - 2s/epoch - 17ms/step\n","Epoch 117/150\n","118/118 - 2s - loss: 0.1680 - accuracy: 0.9371 - val_loss: 0.4542 - val_accuracy: 0.8595 - 2s/epoch - 17ms/step\n","Epoch 118/150\n","118/118 - 2s - loss: 0.1608 - accuracy: 0.9400 - val_loss: 0.4778 - val_accuracy: 0.8548 - 2s/epoch - 17ms/step\n","Epoch 119/150\n","118/118 - 2s - loss: 0.1592 - accuracy: 0.9406 - val_loss: 0.4523 - val_accuracy: 0.8693 - 2s/epoch - 17ms/step\n","Epoch 120/150\n","118/118 - 2s - loss: 0.1531 - accuracy: 0.9437 - val_loss: 0.4849 - val_accuracy: 0.8671 - 2s/epoch - 17ms/step\n","Epoch 121/150\n","118/118 - 2s - loss: 0.1519 - accuracy: 0.9432 - val_loss: 0.4485 - val_accuracy: 0.8687 - 2s/epoch - 17ms/step\n","Epoch 122/150\n","118/118 - 2s - loss: 0.1583 - accuracy: 0.9417 - val_loss: 0.4838 - val_accuracy: 0.8693 - 2s/epoch - 17ms/step\n","Epoch 123/150\n","118/118 - 2s - loss: 0.1460 - accuracy: 0.9448 - val_loss: 0.4579 - val_accuracy: 0.8706 - 2s/epoch - 17ms/step\n","Epoch 124/150\n","118/118 - 2s - loss: 0.1706 - accuracy: 0.9408 - val_loss: 0.3965 - val_accuracy: 0.8723 - 2s/epoch - 17ms/step\n","Epoch 125/150\n","118/118 - 2s - loss: 0.1433 - accuracy: 0.9466 - val_loss: 0.5107 - val_accuracy: 0.8456 - 2s/epoch - 17ms/step\n","Epoch 126/150\n","118/118 - 2s - loss: 0.1429 - accuracy: 0.9480 - val_loss: 0.6594 - val_accuracy: 0.8232 - 2s/epoch - 17ms/step\n","Epoch 127/150\n","118/118 - 2s - loss: 0.1398 - accuracy: 0.9478 - val_loss: 0.4795 - val_accuracy: 0.8513 - 2s/epoch - 17ms/step\n","Epoch 128/150\n","118/118 - 2s - loss: 0.1397 - accuracy: 0.9480 - val_loss: 0.4852 - val_accuracy: 0.8667 - 2s/epoch - 17ms/step\n","Epoch 129/150\n","118/118 - 2s - loss: 0.1335 - accuracy: 0.9507 - val_loss: 0.7514 - val_accuracy: 0.7893 - 2s/epoch - 17ms/step\n","Epoch 130/150\n","118/118 - 2s - loss: 0.1383 - accuracy: 0.9485 - val_loss: 0.5700 - val_accuracy: 0.8310 - 2s/epoch - 17ms/step\n","Epoch 131/150\n","118/118 - 2s - loss: 0.1311 - accuracy: 0.9517 - val_loss: 0.4304 - val_accuracy: 0.8745 - 2s/epoch - 17ms/step\n","Epoch 132/150\n","118/118 - 2s - loss: 0.1280 - accuracy: 0.9519 - val_loss: 0.5424 - val_accuracy: 0.8567 - 2s/epoch - 17ms/step\n","Epoch 133/150\n","118/118 - 2s - loss: 0.1285 - accuracy: 0.9523 - val_loss: 0.4990 - val_accuracy: 0.8735 - 2s/epoch - 17ms/step\n","Epoch 134/150\n","118/118 - 2s - loss: 0.1537 - accuracy: 0.9456 - val_loss: 0.5824 - val_accuracy: 0.8502 - 2s/epoch - 17ms/step\n","Epoch 135/150\n","118/118 - 2s - loss: 0.1287 - accuracy: 0.9530 - val_loss: 0.4983 - val_accuracy: 0.8632 - 2s/epoch - 17ms/step\n","Epoch 136/150\n","118/118 - 2s - loss: 0.1239 - accuracy: 0.9540 - val_loss: 0.5212 - val_accuracy: 0.8692 - 2s/epoch - 17ms/step\n","Epoch 137/150\n","118/118 - 2s - loss: 0.1247 - accuracy: 0.9537 - val_loss: 0.4799 - val_accuracy: 0.8695 - 2s/epoch - 17ms/step\n","Epoch 138/150\n","118/118 - 2s - loss: 0.1184 - accuracy: 0.9557 - val_loss: 0.5252 - val_accuracy: 0.8515 - 2s/epoch - 17ms/step\n","Epoch 139/150\n","118/118 - 2s - loss: 0.1236 - accuracy: 0.9544 - val_loss: 0.5518 - val_accuracy: 0.8539 - 2s/epoch - 17ms/step\n","Epoch 140/150\n","118/118 - 2s - loss: 0.1248 - accuracy: 0.9538 - val_loss: 1.0645 - val_accuracy: 0.8260 - 2s/epoch - 17ms/step\n","Epoch 141/150\n","118/118 - 2s - loss: 0.1383 - accuracy: 0.9513 - val_loss: 0.4735 - val_accuracy: 0.8774 - 2s/epoch - 17ms/step\n","Epoch 142/150\n","118/118 - 2s - loss: 0.1123 - accuracy: 0.9592 - val_loss: 0.7124 - val_accuracy: 0.8174 - 2s/epoch - 17ms/step\n","Epoch 143/150\n","118/118 - 2s - loss: 0.1193 - accuracy: 0.9565 - val_loss: 0.5674 - val_accuracy: 0.8545 - 2s/epoch - 17ms/step\n","Epoch 144/150\n","118/118 - 2s - loss: 0.1157 - accuracy: 0.9577 - val_loss: 0.5344 - val_accuracy: 0.8583 - 2s/epoch - 17ms/step\n","Epoch 145/150\n","118/118 - 2s - loss: 0.1098 - accuracy: 0.9601 - val_loss: 0.5896 - val_accuracy: 0.8628 - 2s/epoch - 17ms/step\n","Epoch 146/150\n","118/118 - 2s - loss: 0.1150 - accuracy: 0.9588 - val_loss: 0.4716 - val_accuracy: 0.8736 - 2s/epoch - 17ms/step\n","Epoch 147/150\n","118/118 - 2s - loss: 0.1112 - accuracy: 0.9590 - val_loss: 0.5509 - val_accuracy: 0.8576 - 2s/epoch - 17ms/step\n","Epoch 148/150\n","118/118 - 2s - loss: 0.1130 - accuracy: 0.9582 - val_loss: 0.4781 - val_accuracy: 0.8743 - 2s/epoch - 17ms/step\n","Epoch 149/150\n","118/118 - 2s - loss: 0.1101 - accuracy: 0.9595 - val_loss: 0.5197 - val_accuracy: 0.8653 - 2s/epoch - 17ms/step\n","Epoch 150/150\n","118/118 - 2s - loss: 0.1140 - accuracy: 0.9592 - val_loss: 0.5163 - val_accuracy: 0.8784 - 2s/epoch - 17ms/step\n"]}],"source":["# solution 2: BetteActivationFunction\n","\n","initializer = tf.keras.initializers.RandomNormal()\n","\n","#################### YOUR CODE HERE ####################\n","# Các bạn thực hiện code (tùy ý tinh chỉnh) để sau môi \n","# hidden layer sẽ là ReLU activation function\n","model = keras.Sequential()\n","model.add(keras.Input(shape=(784,)))\n","model.add(keras.layers.Dense(128, activation='relu', kernel_initializer=initializer))\n","model.add(keras.layers.Dense(128, activation='relu', kernel_initializer=initializer))\n","model.add(keras.layers.Dense(128, activation='relu', kernel_initializer=initializer))\n","model.add(keras.layers.Dense(128, activation='relu', kernel_initializer=initializer))\n","model.add(keras.layers.Dense(128, activation='relu', kernel_initializer=initializer))\n","model.add(keras.layers.Dense(128, activation='relu', kernel_initializer=initializer))\n","model.add(keras.layers.Dense(128, activation='relu', 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=150, verbose=2)"]},{"cell_type":"code","execution_count":3,"id":"KcFzrFXD6XtX","metadata":{"id":"KcFzrFXD6XtX","executionInfo":{"status":"ok","timestamp":1669890360941,"user_tz":-420,"elapsed":12,"user":{"displayName":"Thanh Hùng","userId":"08335145449228425002"}}},"outputs":[],"source":[]},{"cell_type":"code","execution_count":4,"id":"P7yzFL0q6XtY","metadata":{"colab":{"base_uri":"https://localhost:8080/","height":296},"executionInfo":{"elapsed":11,"status":"ok","timestamp":1669890360941,"user":{"displayName":"Thanh Hùng","userId":"08335145449228425002"},"user_tz":-420},"id":"P7yzFL0q6XtY","outputId":"c17aabfb-d21c-4938-e5da-7294e37c8b16"},"outputs":[{"output_type":"execute_result","data":{"text/plain":["<matplotlib.legend.Legend at 0x7f0cfeb1d340>"]},"metadata":{},"execution_count":4},{"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":4,"id":"EN-uP39u6Xtb","metadata":{"id":"EN-uP39u6Xtb","executionInfo":{"status":"ok","timestamp":1669890361985,"user_tz":-420,"elapsed":1051,"user":{"displayName":"Thanh Hùng","userId":"08335145449228425002"}}},"outputs":[],"source":[]},{"cell_type":"code","execution_count":5,"id":"nIpmjpBL6Xtd","metadata":{"colab":{"base_uri":"https://localhost:8080/","height":296},"executionInfo":{"elapsed":10,"status":"ok","timestamp":1669890361985,"user":{"displayName":"Thanh Hùng","userId":"08335145449228425002"},"user_tz":-420},"id":"nIpmjpBL6Xtd","outputId":"1dd6ef50-a546-459f-e7de-5224d095ef51"},"outputs":[{"output_type":"execute_result","data":{"text/plain":["<matplotlib.legend.Legend at 0x7f0cfe5f5c40>"]},"metadata":{},"execution_count":5},{"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":5,"id":"916ab79c","metadata":{"id":"916ab79c","executionInfo":{"status":"ok","timestamp":1669890361986,"user_tz":-420,"elapsed":7,"user":{"displayName":"Thanh Hùng","userId":"08335145449228425002"}}},"outputs":[],"source":[]},{"cell_type":"markdown","source":["## (Optional) Better Activation (lr=0.05)"],"metadata":{"id":"57sl4k6A6Y8l"},"id":"57sl4k6A6Y8l"},{"cell_type":"code","execution_count":6,"id":"d6df3583","metadata":{"colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"elapsed":306493,"status":"ok","timestamp":1669890668472,"user":{"displayName":"Thanh Hùng","userId":"08335145449228425002"},"user_tz":-420},"id":"d6df3583","outputId":"6733d73f-04ba-4960-9311-83b06195132a"},"outputs":[{"output_type":"stream","name":"stdout","text":["Epoch 1/150\n","118/118 - 3s - loss: 1.2138 - accuracy: 0.5882 - val_loss: 0.9556 - val_accuracy: 0.6603 - 3s/epoch - 26ms/step\n","Epoch 2/150\n","118/118 - 2s - loss: 0.7443 - accuracy: 0.7289 - val_loss: 0.7080 - val_accuracy: 0.7533 - 2s/epoch - 18ms/step\n","Epoch 3/150\n","118/118 - 2s - loss: 0.6348 - accuracy: 0.7725 - val_loss: 0.7268 - val_accuracy: 0.7251 - 2s/epoch - 17ms/step\n","Epoch 4/150\n","118/118 - 2s - loss: 0.5732 - accuracy: 0.7946 - val_loss: 0.5934 - val_accuracy: 0.7874 - 2s/epoch - 17ms/step\n","Epoch 5/150\n","118/118 - 2s - loss: 0.5406 - accuracy: 0.8066 - val_loss: 0.6296 - val_accuracy: 0.7755 - 2s/epoch - 17ms/step\n","Epoch 6/150\n","118/118 - 2s - loss: 0.5163 - accuracy: 0.8163 - val_loss: 0.6226 - val_accuracy: 0.7813 - 2s/epoch - 17ms/step\n","Epoch 7/150\n","118/118 - 2s - loss: 0.4929 - accuracy: 0.8251 - val_loss: 0.5625 - val_accuracy: 0.7891 - 2s/epoch - 17ms/step\n","Epoch 8/150\n","118/118 - 2s - loss: 0.4715 - accuracy: 0.8324 - val_loss: 0.5393 - val_accuracy: 0.8078 - 2s/epoch - 17ms/step\n","Epoch 9/150\n","118/118 - 2s - loss: 0.4566 - accuracy: 0.8383 - val_loss: 0.5819 - val_accuracy: 0.7889 - 2s/epoch - 17ms/step\n","Epoch 10/150\n","118/118 - 2s - loss: 0.4494 - accuracy: 0.8406 - val_loss: 0.5538 - val_accuracy: 0.8102 - 2s/epoch - 17ms/step\n","Epoch 11/150\n","118/118 - 2s - loss: 0.4343 - accuracy: 0.8450 - val_loss: 0.7287 - val_accuracy: 0.7203 - 2s/epoch - 17ms/step\n","Epoch 12/150\n","118/118 - 2s - loss: 0.4292 - accuracy: 0.8475 - val_loss: 0.5442 - val_accuracy: 0.8045 - 2s/epoch - 17ms/step\n","Epoch 13/150\n","118/118 - 2s - loss: 0.4224 - accuracy: 0.8490 - val_loss: 0.4728 - val_accuracy: 0.8318 - 2s/epoch - 17ms/step\n","Epoch 14/150\n","118/118 - 2s - loss: 0.4112 - accuracy: 0.8537 - val_loss: 0.5039 - val_accuracy: 0.8158 - 2s/epoch - 17ms/step\n","Epoch 15/150\n","118/118 - 2s - loss: 0.4072 - accuracy: 0.8548 - val_loss: 0.4862 - val_accuracy: 0.8247 - 2s/epoch - 17ms/step\n","Epoch 16/150\n","118/118 - 2s - loss: 0.3991 - accuracy: 0.8575 - val_loss: 0.5331 - val_accuracy: 0.8123 - 2s/epoch - 18ms/step\n","Epoch 17/150\n","118/118 - 2s - loss: 0.3930 - accuracy: 0.8593 - val_loss: 0.4533 - val_accuracy: 0.8329 - 2s/epoch - 17ms/step\n","Epoch 18/150\n","118/118 - 2s - loss: 0.3862 - accuracy: 0.8622 - val_loss: 0.4470 - val_accuracy: 0.8405 - 2s/epoch - 17ms/step\n","Epoch 19/150\n","118/118 - 2s - loss: 0.3829 - accuracy: 0.8637 - val_loss: 0.4785 - val_accuracy: 0.8271 - 2s/epoch - 17ms/step\n","Epoch 20/150\n","118/118 - 2s - loss: 0.3781 - accuracy: 0.8651 - val_loss: 0.5259 - val_accuracy: 0.8014 - 2s/epoch - 17ms/step\n","Epoch 21/150\n","118/118 - 2s - loss: 0.3736 - accuracy: 0.8666 - val_loss: 0.4601 - val_accuracy: 0.8318 - 2s/epoch - 17ms/step\n","Epoch 22/150\n","118/118 - 2s - loss: 0.3701 - accuracy: 0.8671 - val_loss: 0.5032 - val_accuracy: 0.8197 - 2s/epoch - 17ms/step\n","Epoch 23/150\n","118/118 - 2s - loss: 0.3663 - accuracy: 0.8687 - val_loss: 0.5438 - val_accuracy: 0.7992 - 2s/epoch - 17ms/step\n","Epoch 24/150\n","118/118 - 2s - loss: 0.3642 - accuracy: 0.8696 - val_loss: 0.4166 - val_accuracy: 0.8484 - 2s/epoch - 17ms/step\n","Epoch 25/150\n","118/118 - 2s - loss: 0.3580 - accuracy: 0.8728 - val_loss: 0.4067 - val_accuracy: 0.8526 - 2s/epoch - 17ms/step\n","Epoch 26/150\n","118/118 - 2s - loss: 0.3531 - accuracy: 0.8735 - val_loss: 0.4348 - val_accuracy: 0.8443 - 2s/epoch - 17ms/step\n","Epoch 27/150\n","118/118 - 2s - loss: 0.3505 - accuracy: 0.8747 - val_loss: 0.4571 - val_accuracy: 0.8327 - 2s/epoch - 17ms/step\n","Epoch 28/150\n","118/118 - 2s - loss: 0.3476 - accuracy: 0.8751 - val_loss: 0.5095 - val_accuracy: 0.8096 - 2s/epoch - 17ms/step\n","Epoch 29/150\n","118/118 - 2s - loss: 0.3466 - accuracy: 0.8763 - val_loss: 0.4394 - val_accuracy: 0.8381 - 2s/epoch - 17ms/step\n","Epoch 30/150\n","118/118 - 2s - loss: 0.3432 - accuracy: 0.8765 - val_loss: 0.4574 - val_accuracy: 0.8281 - 2s/epoch - 17ms/step\n","Epoch 31/150\n","118/118 - 2s - loss: 0.3392 - accuracy: 0.8782 - val_loss: 0.4223 - val_accuracy: 0.8470 - 2s/epoch - 17ms/step\n","Epoch 32/150\n","118/118 - 2s - loss: 0.3351 - accuracy: 0.8794 - val_loss: 0.4341 - val_accuracy: 0.8440 - 2s/epoch - 17ms/step\n","Epoch 33/150\n","118/118 - 2s - loss: 0.3355 - accuracy: 0.8788 - val_loss: 0.4250 - val_accuracy: 0.8479 - 2s/epoch - 17ms/step\n","Epoch 34/150\n","118/118 - 2s - loss: 0.3309 - accuracy: 0.8814 - val_loss: 0.4010 - val_accuracy: 0.8542 - 2s/epoch - 17ms/step\n","Epoch 35/150\n","118/118 - 2s - loss: 0.3282 - accuracy: 0.8820 - val_loss: 0.4316 - val_accuracy: 0.8381 - 2s/epoch - 17ms/step\n","Epoch 36/150\n","118/118 - 2s - loss: 0.3276 - accuracy: 0.8824 - val_loss: 0.4232 - val_accuracy: 0.8431 - 2s/epoch - 17ms/step\n","Epoch 37/150\n","118/118 - 2s - loss: 0.3247 - accuracy: 0.8820 - val_loss: 0.4040 - val_accuracy: 0.8548 - 2s/epoch - 17ms/step\n","Epoch 38/150\n","118/118 - 2s - loss: 0.3200 - accuracy: 0.8840 - val_loss: 0.3839 - val_accuracy: 0.8614 - 2s/epoch - 17ms/step\n","Epoch 39/150\n","118/118 - 2s - loss: 0.3198 - accuracy: 0.8834 - val_loss: 0.4065 - val_accuracy: 0.8546 - 2s/epoch - 17ms/step\n","Epoch 40/150\n","118/118 - 2s - loss: 0.3178 - accuracy: 0.8855 - val_loss: 0.4271 - val_accuracy: 0.8467 - 2s/epoch - 17ms/step\n","Epoch 41/150\n","118/118 - 2s - loss: 0.3160 - accuracy: 0.8856 - val_loss: 0.3934 - val_accuracy: 0.8577 - 2s/epoch - 17ms/step\n","Epoch 42/150\n","118/118 - 2s - loss: 0.3132 - accuracy: 0.8868 - val_loss: 0.7031 - val_accuracy: 0.7702 - 2s/epoch - 17ms/step\n","Epoch 43/150\n","118/118 - 2s - loss: 0.3171 - accuracy: 0.8860 - val_loss: 0.4310 - val_accuracy: 0.8437 - 2s/epoch - 17ms/step\n","Epoch 44/150\n","118/118 - 2s - loss: 0.3082 - accuracy: 0.8888 - val_loss: 0.3972 - val_accuracy: 0.8570 - 2s/epoch - 17ms/step\n","Epoch 45/150\n","118/118 - 2s - loss: 0.3053 - accuracy: 0.8896 - val_loss: 0.3864 - val_accuracy: 0.8601 - 2s/epoch - 17ms/step\n","Epoch 46/150\n","118/118 - 2s - loss: 0.3021 - accuracy: 0.8912 - val_loss: 0.4569 - val_accuracy: 0.8311 - 2s/epoch - 17ms/step\n","Epoch 47/150\n","118/118 - 2s - loss: 0.3027 - accuracy: 0.8893 - val_loss: 0.4384 - val_accuracy: 0.8453 - 2s/epoch - 17ms/step\n","Epoch 48/150\n","118/118 - 2s - loss: 0.3005 - accuracy: 0.8903 - val_loss: 0.4279 - val_accuracy: 0.8500 - 2s/epoch - 17ms/step\n","Epoch 49/150\n","118/118 - 2s - loss: 0.2962 - accuracy: 0.8925 - val_loss: 0.4351 - val_accuracy: 0.8412 - 2s/epoch - 17ms/step\n","Epoch 50/150\n","118/118 - 2s - loss: 0.2946 - accuracy: 0.8932 - val_loss: 0.3963 - val_accuracy: 0.8550 - 2s/epoch - 17ms/step\n","Epoch 51/150\n","118/118 - 2s - loss: 0.2940 - accuracy: 0.8922 - val_loss: 0.3864 - val_accuracy: 0.8599 - 2s/epoch - 17ms/step\n","Epoch 52/150\n","118/118 - 2s - loss: 0.2918 - accuracy: 0.8939 - val_loss: 0.4023 - val_accuracy: 0.8576 - 2s/epoch - 17ms/step\n","Epoch 53/150\n","118/118 - 2s - loss: 0.2916 - accuracy: 0.8931 - val_loss: 0.4128 - val_accuracy: 0.8484 - 2s/epoch - 17ms/step\n","Epoch 54/150\n","118/118 - 2s - loss: 0.2922 - accuracy: 0.8935 - val_loss: 0.4110 - val_accuracy: 0.8537 - 2s/epoch - 17ms/step\n","Epoch 55/150\n","118/118 - 2s - loss: 0.2887 - accuracy: 0.8942 - val_loss: 0.4302 - val_accuracy: 0.8464 - 2s/epoch - 17ms/step\n","Epoch 56/150\n","118/118 - 2s - loss: 0.2845 - accuracy: 0.8960 - val_loss: 0.3976 - val_accuracy: 0.8620 - 2s/epoch - 17ms/step\n","Epoch 57/150\n","118/118 - 2s - loss: 0.2819 - accuracy: 0.8972 - val_loss: 0.4486 - val_accuracy: 0.8409 - 2s/epoch - 17ms/step\n","Epoch 58/150\n","118/118 - 2s - loss: 0.2835 - accuracy: 0.8964 - val_loss: 0.4045 - val_accuracy: 0.8542 - 2s/epoch - 17ms/step\n","Epoch 59/150\n","118/118 - 2s - loss: 0.2829 - accuracy: 0.8965 - val_loss: 0.4111 - val_accuracy: 0.8572 - 2s/epoch - 18ms/step\n","Epoch 60/150\n","118/118 - 2s - loss: 0.2795 - accuracy: 0.8978 - val_loss: 0.4181 - val_accuracy: 0.8515 - 2s/epoch - 18ms/step\n","Epoch 61/150\n","118/118 - 2s - loss: 0.2768 - accuracy: 0.8986 - val_loss: 0.3728 - val_accuracy: 0.8687 - 2s/epoch - 17ms/step\n","Epoch 62/150\n","118/118 - 2s - loss: 0.2764 - accuracy: 0.8985 - val_loss: 0.4140 - val_accuracy: 0.8549 - 2s/epoch - 17ms/step\n","Epoch 63/150\n","118/118 - 2s - loss: 0.2760 - accuracy: 0.8997 - val_loss: 0.4087 - val_accuracy: 0.8553 - 2s/epoch - 17ms/step\n","Epoch 64/150\n","118/118 - 2s - loss: 0.2720 - accuracy: 0.8999 - val_loss: 0.3714 - val_accuracy: 0.8662 - 2s/epoch - 17ms/step\n","Epoch 65/150\n","118/118 - 2s - loss: 0.2705 - accuracy: 0.9002 - val_loss: 0.3985 - val_accuracy: 0.8607 - 2s/epoch - 17ms/step\n","Epoch 66/150\n","118/118 - 2s - loss: 0.2704 - accuracy: 0.9017 - val_loss: 0.5196 - val_accuracy: 0.8090 - 2s/epoch - 17ms/step\n","Epoch 67/150\n","118/118 - 2s - loss: 0.2708 - accuracy: 0.9007 - val_loss: 0.3726 - val_accuracy: 0.8679 - 2s/epoch - 17ms/step\n","Epoch 68/150\n","118/118 - 2s - loss: 0.2688 - accuracy: 0.9015 - val_loss: 0.3931 - val_accuracy: 0.8609 - 2s/epoch - 17ms/step\n","Epoch 69/150\n","118/118 - 2s - loss: 0.2658 - accuracy: 0.9026 - val_loss: 0.3694 - val_accuracy: 0.8704 - 2s/epoch - 17ms/step\n","Epoch 70/150\n","118/118 - 2s - loss: 0.2646 - accuracy: 0.9030 - val_loss: 0.4071 - val_accuracy: 0.8590 - 2s/epoch - 18ms/step\n","Epoch 71/150\n","118/118 - 2s - loss: 0.2632 - accuracy: 0.9040 - val_loss: 0.4093 - val_accuracy: 0.8492 - 2s/epoch - 17ms/step\n","Epoch 72/150\n","118/118 - 2s - loss: 0.2617 - accuracy: 0.9046 - val_loss: 0.4459 - val_accuracy: 0.8498 - 2s/epoch - 17ms/step\n","Epoch 73/150\n","118/118 - 2s - loss: 0.2601 - accuracy: 0.9051 - val_loss: 0.4008 - val_accuracy: 0.8591 - 2s/epoch - 17ms/step\n","Epoch 74/150\n","118/118 - 2s - loss: 0.2608 - accuracy: 0.9042 - val_loss: 0.3789 - val_accuracy: 0.8677 - 2s/epoch - 17ms/step\n","Epoch 75/150\n","118/118 - 2s - loss: 0.2567 - accuracy: 0.9050 - val_loss: 0.3741 - val_accuracy: 0.8694 - 2s/epoch - 17ms/step\n","Epoch 76/150\n","118/118 - 2s - loss: 0.2553 - accuracy: 0.9064 - val_loss: 0.4286 - val_accuracy: 0.8579 - 2s/epoch - 17ms/step\n","Epoch 77/150\n","118/118 - 2s - loss: 0.2553 - accuracy: 0.9063 - val_loss: 0.4042 - val_accuracy: 0.8604 - 2s/epoch - 17ms/step\n","Epoch 78/150\n","118/118 - 2s - loss: 0.2520 - accuracy: 0.9072 - val_loss: 0.4421 - val_accuracy: 0.8476 - 2s/epoch - 17ms/step\n","Epoch 79/150\n","118/118 - 2s - loss: 0.2535 - accuracy: 0.9068 - val_loss: 0.4646 - val_accuracy: 0.8274 - 2s/epoch - 17ms/step\n","Epoch 80/150\n","118/118 - 2s - loss: 0.2514 - accuracy: 0.9069 - val_loss: 0.3858 - val_accuracy: 0.8586 - 2s/epoch - 17ms/step\n","Epoch 81/150\n","118/118 - 2s - loss: 0.2489 - accuracy: 0.9084 - val_loss: 0.4305 - val_accuracy: 0.8545 - 2s/epoch - 17ms/step\n","Epoch 82/150\n","118/118 - 2s - loss: 0.2492 - accuracy: 0.9086 - val_loss: 0.4056 - val_accuracy: 0.8610 - 2s/epoch - 17ms/step\n","Epoch 83/150\n","118/118 - 2s - loss: 0.2484 - accuracy: 0.9085 - val_loss: 0.3917 - val_accuracy: 0.8604 - 2s/epoch - 17ms/step\n","Epoch 84/150\n","118/118 - 2s - loss: 0.2454 - accuracy: 0.9102 - val_loss: 0.4124 - val_accuracy: 0.8546 - 2s/epoch - 17ms/step\n","Epoch 85/150\n","118/118 - 2s - loss: 0.2452 - accuracy: 0.9091 - val_loss: 0.4564 - val_accuracy: 0.8378 - 2s/epoch - 17ms/step\n","Epoch 86/150\n","118/118 - 2s - loss: 0.2450 - accuracy: 0.9111 - val_loss: 0.3920 - val_accuracy: 0.8657 - 2s/epoch - 17ms/step\n","Epoch 87/150\n","118/118 - 2s - loss: 0.2419 - accuracy: 0.9117 - val_loss: 0.4052 - val_accuracy: 0.8617 - 2s/epoch - 17ms/step\n","Epoch 88/150\n","118/118 - 2s - loss: 0.2392 - accuracy: 0.9125 - val_loss: 0.4051 - val_accuracy: 0.8635 - 2s/epoch - 17ms/step\n","Epoch 89/150\n","118/118 - 2s - loss: 0.2400 - accuracy: 0.9116 - val_loss: 0.3809 - val_accuracy: 0.8654 - 2s/epoch - 17ms/step\n","Epoch 90/150\n","118/118 - 2s - loss: 0.2387 - accuracy: 0.9123 - val_loss: 0.4153 - val_accuracy: 0.8568 - 2s/epoch - 17ms/step\n","Epoch 91/150\n","118/118 - 2s - loss: 0.2363 - accuracy: 0.9126 - val_loss: 0.4169 - val_accuracy: 0.8583 - 2s/epoch - 17ms/step\n","Epoch 92/150\n","118/118 - 2s - loss: 0.2344 - accuracy: 0.9140 - val_loss: 0.4283 - val_accuracy: 0.8545 - 2s/epoch - 17ms/step\n","Epoch 93/150\n","118/118 - 2s - loss: 0.2350 - accuracy: 0.9142 - val_loss: 0.3972 - val_accuracy: 0.8636 - 2s/epoch - 17ms/step\n","Epoch 94/150\n","118/118 - 2s - loss: 0.2336 - accuracy: 0.9146 - val_loss: 0.4212 - val_accuracy: 0.8583 - 2s/epoch - 17ms/step\n","Epoch 95/150\n","118/118 - 2s - loss: 0.2302 - accuracy: 0.9157 - val_loss: 0.3798 - val_accuracy: 0.8716 - 2s/epoch - 17ms/step\n","Epoch 96/150\n","118/118 - 2s - loss: 0.2307 - accuracy: 0.9157 - val_loss: 0.4340 - val_accuracy: 0.8482 - 2s/epoch - 17ms/step\n","Epoch 97/150\n","118/118 - 2s - loss: 0.2319 - accuracy: 0.9144 - val_loss: 0.4725 - val_accuracy: 0.8377 - 2s/epoch - 17ms/step\n","Epoch 98/150\n","118/118 - 2s - loss: 0.2283 - accuracy: 0.9167 - val_loss: 0.5458 - val_accuracy: 0.8239 - 2s/epoch - 17ms/step\n","Epoch 99/150\n","118/118 - 2s - loss: 0.2422 - accuracy: 0.9128 - val_loss: 0.4321 - val_accuracy: 0.8467 - 2s/epoch - 17ms/step\n","Epoch 100/150\n","118/118 - 2s - loss: 0.2261 - accuracy: 0.9167 - val_loss: 0.3911 - val_accuracy: 0.8684 - 2s/epoch - 17ms/step\n","Epoch 101/150\n","118/118 - 2s - loss: 0.2251 - accuracy: 0.9175 - val_loss: 0.3901 - val_accuracy: 0.8674 - 2s/epoch - 17ms/step\n","Epoch 102/150\n","118/118 - 2s - loss: 0.2226 - accuracy: 0.9189 - val_loss: 0.3852 - val_accuracy: 0.8659 - 2s/epoch - 17ms/step\n","Epoch 103/150\n","118/118 - 2s - loss: 0.2235 - accuracy: 0.9176 - val_loss: 0.3952 - val_accuracy: 0.8611 - 2s/epoch - 17ms/step\n","Epoch 104/150\n","118/118 - 2s - loss: 0.2211 - accuracy: 0.9194 - val_loss: 0.4032 - val_accuracy: 0.8606 - 2s/epoch - 17ms/step\n","Epoch 105/150\n","118/118 - 2s - loss: 0.2227 - accuracy: 0.9194 - val_loss: 0.3968 - val_accuracy: 0.8653 - 2s/epoch - 17ms/step\n","Epoch 106/150\n","118/118 - 3s - loss: 0.2218 - accuracy: 0.9188 - val_loss: 0.3980 - val_accuracy: 0.8655 - 3s/epoch - 24ms/step\n","Epoch 107/150\n","118/118 - 2s - loss: 0.2178 - accuracy: 0.9192 - val_loss: 0.4084 - val_accuracy: 0.8644 - 2s/epoch - 17ms/step\n","Epoch 108/150\n","118/118 - 2s - loss: 0.2181 - accuracy: 0.9208 - val_loss: 0.3825 - val_accuracy: 0.8718 - 2s/epoch - 17ms/step\n","Epoch 109/150\n","118/118 - 2s - loss: 0.2157 - accuracy: 0.9208 - val_loss: 0.4767 - val_accuracy: 0.8412 - 2s/epoch - 17ms/step\n","Epoch 110/150\n","118/118 - 2s - loss: 0.2161 - accuracy: 0.9203 - val_loss: 0.4003 - val_accuracy: 0.8650 - 2s/epoch - 17ms/step\n","Epoch 111/150\n","118/118 - 2s - loss: 0.2157 - accuracy: 0.9210 - val_loss: 0.4111 - val_accuracy: 0.8628 - 2s/epoch - 17ms/step\n","Epoch 112/150\n","118/118 - 2s - loss: 0.2152 - accuracy: 0.9202 - val_loss: 0.4293 - val_accuracy: 0.8588 - 2s/epoch - 17ms/step\n","Epoch 113/150\n","118/118 - 2s - loss: 0.2106 - accuracy: 0.9223 - val_loss: 0.4274 - val_accuracy: 0.8550 - 2s/epoch - 17ms/step\n","Epoch 114/150\n","118/118 - 2s - loss: 0.2108 - accuracy: 0.9218 - val_loss: 0.4138 - val_accuracy: 0.8627 - 2s/epoch - 17ms/step\n","Epoch 115/150\n","118/118 - 2s - loss: 0.2105 - accuracy: 0.9221 - val_loss: 0.4368 - val_accuracy: 0.8513 - 2s/epoch - 17ms/step\n","Epoch 116/150\n","118/118 - 2s - loss: 0.2106 - accuracy: 0.9218 - val_loss: 0.4126 - val_accuracy: 0.8609 - 2s/epoch - 17ms/step\n","Epoch 117/150\n","118/118 - 2s - loss: 0.2078 - accuracy: 0.9236 - val_loss: 0.4955 - val_accuracy: 0.8365 - 2s/epoch - 17ms/step\n","Epoch 118/150\n","118/118 - 2s - loss: 0.2086 - accuracy: 0.9226 - val_loss: 0.3980 - val_accuracy: 0.8663 - 2s/epoch - 17ms/step\n","Epoch 119/150\n","118/118 - 2s - loss: 0.2050 - accuracy: 0.9237 - val_loss: 0.4255 - val_accuracy: 0.8656 - 2s/epoch - 17ms/step\n","Epoch 120/150\n","118/118 - 2s - loss: 0.2078 - accuracy: 0.9235 - val_loss: 0.3877 - val_accuracy: 0.8705 - 2s/epoch - 17ms/step\n","Epoch 121/150\n","118/118 - 2s - loss: 0.2029 - accuracy: 0.9258 - val_loss: 0.4270 - val_accuracy: 0.8610 - 2s/epoch - 17ms/step\n","Epoch 122/150\n","118/118 - 2s - loss: 0.2020 - accuracy: 0.9258 - val_loss: 0.4125 - val_accuracy: 0.8674 - 2s/epoch - 17ms/step\n","Epoch 123/150\n","118/118 - 2s - loss: 0.2031 - accuracy: 0.9240 - val_loss: 0.4147 - val_accuracy: 0.8687 - 2s/epoch - 17ms/step\n","Epoch 124/150\n","118/118 - 2s - loss: 0.2020 - accuracy: 0.9263 - val_loss: 0.3784 - val_accuracy: 0.8738 - 2s/epoch - 17ms/step\n","Epoch 125/150\n","118/118 - 2s - loss: 0.2030 - accuracy: 0.9247 - val_loss: 0.4176 - val_accuracy: 0.8625 - 2s/epoch - 17ms/step\n","Epoch 126/150\n","118/118 - 2s - loss: 0.1992 - accuracy: 0.9272 - val_loss: 0.4165 - val_accuracy: 0.8595 - 2s/epoch - 17ms/step\n","Epoch 127/150\n","118/118 - 2s - loss: 0.1994 - accuracy: 0.9271 - val_loss: 0.4856 - val_accuracy: 0.8496 - 2s/epoch - 17ms/step\n","Epoch 128/150\n","118/118 - 2s - loss: 0.1977 - accuracy: 0.9266 - val_loss: 0.4746 - val_accuracy: 0.8474 - 2s/epoch - 17ms/step\n","Epoch 129/150\n","118/118 - 2s - loss: 0.1978 - accuracy: 0.9276 - val_loss: 0.4284 - val_accuracy: 0.8543 - 2s/epoch - 17ms/step\n","Epoch 130/150\n","118/118 - 2s - loss: 0.1948 - accuracy: 0.9286 - val_loss: 0.4059 - val_accuracy: 0.8650 - 2s/epoch - 17ms/step\n","Epoch 131/150\n","118/118 - 2s - loss: 0.1953 - accuracy: 0.9281 - val_loss: 0.3900 - val_accuracy: 0.8719 - 2s/epoch - 17ms/step\n","Epoch 132/150\n","118/118 - 2s - loss: 0.1923 - accuracy: 0.9289 - val_loss: 0.5509 - val_accuracy: 0.8310 - 2s/epoch - 17ms/step\n","Epoch 133/150\n","118/118 - 2s - loss: 0.1946 - accuracy: 0.9280 - val_loss: 0.5007 - val_accuracy: 0.8496 - 2s/epoch - 17ms/step\n","Epoch 134/150\n","118/118 - 2s - loss: 0.1930 - accuracy: 0.9289 - val_loss: 0.4886 - val_accuracy: 0.8469 - 2s/epoch - 17ms/step\n","Epoch 135/150\n","118/118 - 2s - loss: 0.1920 - accuracy: 0.9300 - val_loss: 0.4378 - val_accuracy: 0.8562 - 2s/epoch - 17ms/step\n","Epoch 136/150\n","118/118 - 2s - loss: 0.1879 - accuracy: 0.9311 - val_loss: 0.4275 - val_accuracy: 0.8624 - 2s/epoch - 17ms/step\n","Epoch 137/150\n","118/118 - 2s - loss: 0.1878 - accuracy: 0.9305 - val_loss: 0.4523 - val_accuracy: 0.8627 - 2s/epoch - 17ms/step\n","Epoch 138/150\n","118/118 - 2s - loss: 0.1876 - accuracy: 0.9312 - val_loss: 0.4428 - val_accuracy: 0.8481 - 2s/epoch - 17ms/step\n","Epoch 139/150\n","118/118 - 2s - loss: 0.1901 - accuracy: 0.9298 - val_loss: 0.3962 - val_accuracy: 0.8747 - 2s/epoch - 17ms/step\n","Epoch 140/150\n","118/118 - 2s - loss: 0.1913 - accuracy: 0.9298 - val_loss: 0.4211 - val_accuracy: 0.8604 - 2s/epoch - 17ms/step\n","Epoch 141/150\n","118/118 - 2s - loss: 0.1854 - accuracy: 0.9312 - val_loss: 0.5127 - val_accuracy: 0.8369 - 2s/epoch - 17ms/step\n","Epoch 142/150\n","118/118 - 2s - loss: 0.1825 - accuracy: 0.9325 - val_loss: 0.4677 - val_accuracy: 0.8568 - 2s/epoch - 17ms/step\n","Epoch 143/150\n","118/118 - 2s - loss: 0.1842 - accuracy: 0.9321 - val_loss: 0.4565 - val_accuracy: 0.8567 - 2s/epoch - 17ms/step\n","Epoch 144/150\n","118/118 - 2s - loss: 0.1806 - accuracy: 0.9337 - val_loss: 0.4033 - val_accuracy: 0.8676 - 2s/epoch - 17ms/step\n","Epoch 145/150\n","118/118 - 2s - loss: 0.1799 - accuracy: 0.9337 - val_loss: 0.5524 - val_accuracy: 0.8467 - 2s/epoch - 17ms/step\n","Epoch 146/150\n","118/118 - 2s - loss: 0.1796 - accuracy: 0.9354 - val_loss: 0.4337 - val_accuracy: 0.8653 - 2s/epoch - 17ms/step\n","Epoch 147/150\n","118/118 - 2s - loss: 0.1804 - accuracy: 0.9334 - val_loss: 0.4211 - val_accuracy: 0.8672 - 2s/epoch - 17ms/step\n","Epoch 148/150\n","118/118 - 2s - loss: 0.1805 - accuracy: 0.9335 - val_loss: 0.4698 - val_accuracy: 0.8553 - 2s/epoch - 17ms/step\n","Epoch 149/150\n","118/118 - 2s - loss: 0.1790 - accuracy: 0.9336 - val_loss: 0.4502 - val_accuracy: 0.8614 - 2s/epoch - 17ms/step\n","Epoch 150/150\n","118/118 - 2s - loss: 0.1804 - accuracy: 0.9338 - val_loss: 0.4639 - val_accuracy: 0.8597 - 2s/epoch - 17ms/step\n"]}],"source":["# solution 2: BetteActivationFunction reduce LR\n","\n","initializer = tf.keras.initializers.RandomNormal()\n","\n","#################### YOUR CODE HERE ####################\n","# Các bạn thực hiện code (tùy ý tinh chỉnh) để sau môi \n","# hidden layer sẽ là ReLU activation function\n","model = keras.Sequential()\n","model.add(keras.Input(shape=(784,)))\n","model.add(keras.layers.Dense(128, activation='relu', kernel_initializer=initializer))\n","model.add(keras.layers.Dense(128, activation='relu', kernel_initializer=initializer))\n","model.add(keras.layers.Dense(128, activation='relu', kernel_initializer=initializer))\n","model.add(keras.layers.Dense(128, activation='relu', kernel_initializer=initializer))\n","model.add(keras.layers.Dense(128, activation='relu', kernel_initializer=initializer))\n","model.add(keras.layers.Dense(128, activation='relu', kernel_initializer=initializer))\n","model.add(keras.layers.Dense(128, activation='relu', kernel_initializer=initializer))\n","model.add(keras.layers.Dense(10))\n","########################################################\n","\n","\n","# optimizer and loss\n","model.compile(optimizer=tf.keras.optimizers.SGD(0.005), \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=150, verbose=2)"]},{"cell_type":"code","execution_count":6,"id":"4da831e7","metadata":{"id":"4da831e7","executionInfo":{"status":"ok","timestamp":1669890668473,"user_tz":-420,"elapsed":18,"user":{"displayName":"Thanh Hùng","userId":"08335145449228425002"}}},"outputs":[],"source":[]},{"cell_type":"code","execution_count":7,"id":"248dd07e","metadata":{"colab":{"base_uri":"https://localhost:8080/","height":296},"executionInfo":{"elapsed":17,"status":"ok","timestamp":1669890668473,"user":{"displayName":"Thanh Hùng","userId":"08335145449228425002"},"user_tz":-420},"id":"248dd07e","outputId":"a4ae7472-344c-4afa-a7ed-34d3b7dcdb05"},"outputs":[{"output_type":"execute_result","data":{"text/plain":["<matplotlib.legend.Legend at 0x7f0cfc2ab910>"]},"metadata":{},"execution_count":7},{"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":7,"id":"8e35d7eb","metadata":{"id":"8e35d7eb","executionInfo":{"status":"ok","timestamp":1669890668474,"user_tz":-420,"elapsed":14,"user":{"displayName":"Thanh Hùng","userId":"08335145449228425002"}}},"outputs":[],"source":[]},{"cell_type":"code","execution_count":8,"id":"4b997d5b","metadata":{"colab":{"base_uri":"https://localhost:8080/","height":299},"executionInfo":{"elapsed":14,"status":"ok","timestamp":1669890668474,"user":{"displayName":"Thanh Hùng","userId":"08335145449228425002"},"user_tz":-420},"id":"4b997d5b","outputId":"a3d5cc04-23c6-4a74-e20c-ada79b923e3b"},"outputs":[{"output_type":"execute_result","data":{"text/plain":["<matplotlib.legend.Legend at 0x7f0cfd2b5ca0>"]},"metadata":{},"execution_count":8},{"output_type":"display_data","data":{"text/plain":["<Figure size 432x288 with 1 Axes>"],"image/png":"iVBORw0KGgoAAAANSUhEUgAAAYgAAAEJCAYAAACOr7BbAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nOydd5yU1bn4v8/2yrKVDrtILwKKSAQFNSpWFHs0ibFdE6Mx5np/pGqMuXpzTTS5MSZqjNFYorEEI2pAQVFBKQLC0pGytN1le53ZmfP747wvM7vM7s6W2QLP9/PZz8xbzvued3bmPOepR4wxKIqiKEpTorq7A4qiKErPRAWEoiiKEhIVEIqiKEpIVEAoiqIoIVEBoSiKooREBYSiKIoSkogKCBGZIyJbRGS7iMwPcXyYiLwnIutFZKmIDA465hORtc7fgkj2U1EURTkaiVQehIhEA1uBc4ACYCVwrTEmP+icV4B/GWP+KiJnAd8yxnzdOVZljEmJSOcURVGUVomJ4LWnAduNMTsBROQlYC6QH3TOOOBu5/0S4I323iwrK8vk5ua2t7miKMpxyerVq4uNMdmhjkVSQAwC9gZtFwCnNjlnHTAP+C1wGZAqIpnGmMNAgoisAhqAh4wxLQqP3NxcVq1a1WmdVxRFOR4Qkd3NHetuJ/V/ArNE5HNgFrAP8DnHhhljpgJfAx4VkROaNhaRW0VklYisKioq6rJOK4qiHA9EUkDsA4YEbQ929h3BGLPfGDPPGDMF+LGzr8x53ee87gSWAlOa3sAY84QxZqoxZmp2dkgNSVEURWknkRQQK4GRIpInInHANUCjaCQRyRIRtw8/BJ529qeLSLx7DjCDxr4LRVEUJcJETEAYYxqA7wLvApuAl40xG0XkfhG5xDltNrBFRLYC/YBfOvvHAqtEZB3Wef1QcPSToiiKEnkiFuba1UydOtWok1pRFKVtiMhqx997FN3tpFYURVF6KCogFEVRlJCogFAURemh1Hp8vLamgBU7D3fL/SOZKKcoiqK0gaLKel7/vICiynrqG/y8tf4Ah6s9AJw1JodzxvWjzuuj1uujzuun3nk/IC2Rb88+KlWsw6iAUBRF6UR8fkN0lADg9xtKajxkpcQfdd7OoireXHeAxZsOERUl9EmI4dOdJXh8fhJjoxGBaXkZ3Hr6cL7YV87v39/O+5sLG10jITaKxNhoJg/pqwJCURSlp2GMYeP+Ct7bVMjiTYfYsL+cfqkJ9EtLYEdhFVX1DUwY1Ierpg4hJzWBijqvYzYqQQROHppOYlw0xVUerpk2hG+elssJ2Y3rlJ42Iovrpw+jos5LQkw0iXHRxMdEISIRfTYNc1UURWmCz2/4Yl85H2wpYl9ZDUMzkoiNjmJdQRnFlR6mDO3L4PREPt9bxvIdhzlQXocITBnSl1PyMiiqqOdgRR0nZKfQPy2BN9ftZ/PByiPXH5KRyLXThjJvymD6pyV045O2HOaqGoSiKMcc9Q0+quttWbf0pNgjM+3NByuIEmFUv1TKajz898JNrC8oJzY6ir5JsQzNSKKqvoEPtxZRWuNFBDKT4yiusn6AIRmJZKXE85ePd+Hx+clMjmNaXgbfPyeHs8bkhDQlAXxn9gnsOlxDrcdHTLQwIjuFqKjIzv47AxUQiqIcMxhjeGV1Ab98axPltV4ARvdL5c6zR7Ji52GeW2ELl84Ykcm2Q1WUVHs4Y1Q2xhhKqj38a/0BYqOFM0fnMGt0NqePzCYjOY4aTwP1Xj/pyXEA1Hl9FFXWMzg9MSwzj4iQl5UcuQePECogFEXpcfj9JuQM2+c3/Gv9fv6+ci/DMpP5ygmZbDtUybJtxTT4/XgbDFsOVTItN4PzJ/bH6/Pz0sq93P7CGkTgxhl5ZKXG8ewnu8lKiefpG05hwqC0VvuTFBdDUlxgOyE2miEZSZ35yD0S9UEoitIl1Hl9bC+sAmz0zYC0ROJjothZXM2hijpG90/FGPj1v7fw6pp99E2MJadPAtX1DVTWeUmOj8HnNxwor2NoRhIl1R6q6huIEjhpaDppibHUNfiYM2EA100bekTANPj8vLe5kIFpiUwc3LowON5QH4SiKJ2O1+cHIDba5tuu3l3Khn3lREUJaYmxjO6XSt+kWApKa/lgaxHPr9h9JKbfJTpK8PlNo+0ogaumDgaEoso6kuNjSE2IoabeR43Hx48vHMAFEwbQ4DfkH6ggNzOJvsHT+ybEREdx3vj+nf8BHAeogFAUJWxqPT6+2FfOm+v288bafWDgvAn9OVRRx7Jtxc22E4Gzx+Qwd/Ig4mOiqPX62FdWS1VdAyP7pZCTmsDmg5UUVtbxtWlDGZbZur0+LkqYPKRvZz6e0gQVEIpyHFPn9bFyVwm1Hh+x0VGcODiNzJR48vdX8ODbm5g9OoebZuZR5/Vx10trWbTpED6/IT4mivMn9CcmOop3NxwkLiaKH10whksnDwKB4koP2worqaj1Mjg9iVH9UxnUN7HFvswYkdVFT62EiwoIRTkOqPX4+Hh7MQcq6jhcVc/hKg/7y2pZvvMwNR7fkfNiooSThqWzencpAizbVkxcTBQfbSvi3Y2HuGlmHtOHZzItN4O0pFgAHpw3EcGaclxyUhMYN7BPFz+l0tmogFCUXowxhoLSWvIPVFBYUceEQWmM7p/KzqJqthyspKiqnp1FVbz9xUEq6xuOtOubFEtWSjyXThnEOeP6kZ0ST43Hx3ubDrEo/xCXnzSIe84bwz3/WMdP39gAwL0Xj+NbM/KO6kNstNb8PFbRKCZF6QXUN/gor/GS0ycBv9/w+uf7+PvKvWw6UNFo4A9FSnwM547vx+UnDWZkTgrpyXFhD+q1Hh8/eGUtJw7uy22zOr/Wj9L9aBSTovQS6ht87C2poazGi6fBT2xMFJ/vKeXJZV9SVFnP2AF9iI0W1heUM7pfKpdOGcTYAX0YOyCV7NR41u0tZ3thFSfkJDN2QB/690kgOb79P/PEuGj+cN3JnfiESm9CBYSidAF+J5QzKkooKK1h5a4S6rx+BvZNJCkumjqvj4+2F/Pyyr2U1niPaj9zRBY3nJbL+5sLKa6q5+ErJzFvyqCjkskGpx/7yVtK16ECQlE6ibIaD7sP1zCwbyLZqfH4/YZNByt44/N9vLZmH4erPcRFR+Fx8geaEh0lnDO2H+dN6EdGcjzxMVF4fX6yUuIZO8A6fG8/c0RXPpJynBNRASEic4DfAtHAU8aYh5ocHwY8DWQDJcD1xpgC59g3gZ84pz5gjPlrJPuqKOFijKGsxsvHO4r5eHsxWw9V8WVxNSVBSWC5mTbTt6KugZgo4atj+zG6fyp1DT7690ng1LxM0pJi2V9WS53XR0JsNMMyksjp072VPRUlmIgJCBGJBh4DzgEKgJUissAYkx902sPAs8aYv4rIWcCDwNdFJAO4F5gKGGC107Y0Uv1VFJcaTwPrC8oZ278PaUmxVNc3sGp3Ke9tOsRH24rZX15LnddqAakJMYwb0IfzxvdjeFYKQzKS2H24mtW7S8lMiePkYRnMGpVNdmroKp+t5QYoSncSSQ1iGrDdGLMTQEReAuYCwQJiHHC3834J8Ibz/jxgkTGmxGm7CJgDvBjB/irHGfUNPrYdqqK6voGq+gYKK+vJ329NQpVOjZ+hGUnsLa3F5zckxEYx44QszhqTQ3ZqPFNz05k0uG+j+H9FOZaIpIAYBOwN2i4ATm1yzjpgHtYMdRmQKiKZzbQd1PQGInIrcCvA0KFDO63jyrGFz2/YdKCC1btL2VNSQ0y0UFhRz+JNh6isaxwiGhcTxQUT+jNnQn82Hahk04EKLjpxIFNz05k+PJOE2OhuegpF6Xq620n9n8DvReQG4ENgH+BrsUUQxpgngCfA5kFEooNKz8fr8/PRtmIWrNtPRa2Xk4alMywziTqvn437y3lr/QEKK+sBSIyNxm8MSXHRzBnfn9mjc+ibFEtyfAw5qfFkpcQTF2M1gjkTBnTnYylKtxNJAbEPGBK0PdjZdwRjzH6sBoGIpACXG2PKRGQfMLtJ26UR7KvSC6nz+nhl1V4eX7qD/eV1pCXGkpUSx3tBC7vHxURx5uhszp8wgFPyMhiYlhDxdXwV5VghkgJiJTBSRPKwguEa4GvBJ4hIFlBijPEDP8RGNAG8C/y3iKQ72+c6x5VjmLIaDxv2VbD1UCVDMpKYPjyDZduKeWrZTkSEU3IzSIiNYl9pLVsOVbL5QCUen5+pw9K575LxzB6dQ1xMFKXVHgor60mKiyYzJY6kuO5WlBWldxKxX44xpkFEvosd7KOBp40xG0XkfmCVMWYBVkt4UEQM1sR0u9O2RER+gRUyAPe7Dmuld1Pf4GP17lJOHNyXFCfDt77Bxx+X7uSxpdvxNBydIzA8K5m0pFieWraTBr8hJzWeE7JTuGFGLmeOzmH68IxGWkF6ctyRpSEVRWk/WotJiTi1Hh8b95fz0fZi/rZiD8VV9aQnxXLjjDzKar28u/EgBaW1XHTiAL42bSgjclLYXljF8p2HGdUvlQsmDiA6Sqjz+hCB+Bh1FCtKZ6G1mJRuwRjD31bs5hdvbTqiGcwenc3cyQN5/fP9/HrRVuJiojg1L4NfXjaRWaOyj7TN6ZPAaU3WB9AIIkXpWlRAKB3CGENxlYc6r48D5XUs/OIAn+8tY3S/FGq9ft5ct59Zo7K5fvowJg1JIyfVZgpfNmUwe0tqyE6N14FfUXooKiCUNlHjaWBHYTV52ckcLK/l52/mN1pqMj7Grkr27/xDlNV4ufOsEdz11VFHFZUDGJKhheUUpSejAkIJidfnp6C0ll2Hq6n1+BiSnsSmgxU8/O6WIzkFUWLXGvjBOaPon5ZAn8RYZozIIiU+BmMM1R7fEUe0oii9D/31KkfwNPh5dPFW/rX+APvKbHmJpkwe0pcfXjCGfaW11Df4ueG0XDJTjq4zJCIqHBSll6O/4OMUYwxbDlWyOP8QVfU+hmcn88Kne1i7t4yzx+RwyaSB5GYlk5uZRGJcNHtLaomPiWL26GxNNFOU4wQVEMcZPr/hX+v389iS7Ww9VIWIXaje6zOkxsfwh+tO4oKJR5eYGD8wrRt6qyhKd6IC4him1uNjd0k1S7cU8c4Gm2tQWeelvsHPqH4p/PKyCZwzrh/pSXHsPlxDRnIcGZpgpiiKgwqIY4iKOi8bCspZvKmQRZsOsrek9sixSYPTOGdcDslxMUzNTefccf0bRRaNyEnpji4rSteyZwUkZkD2qK69b1UhPHMhXPE09J/YtffuACogjgF2FFVxxwufk3+gArAF6k4fkcVVJw9hSEYSJw9L15BSpXmqiyExHaIinI/iqYGyPZAzJrL3aY7CTfDXS2DgZLjp321rW1sGJTth0Entu/f+z6F4K+T/UwWE0nV8vqeUG59ZSXSUcM95oxk/sA9TczM0gigS+H2w5lk48SqIS+7u3nQOVUXw6ES44H/hpK9H9l4L/xM2vg7z90J0BL6fxsDnz8Hw2dC3yfowDfXw6i3gq4e9n0HlQUjtH/613/gO7FwCP9wHUe1YIKp0l33d9VHb23YjuhRWL6SgtIZHFm1l3h8+5oo/LqdPYiyvfvs0bj9zBLNH56hwaI6SL2H9K21r4wtaUGjTAvjXXbD6mdbbNdTb8x6fCSufats9u5KdS6Gh1s5wW+NQvjXR+BpaP7cppbtg3UvgrYHKA21vHw4rn4IFd8C7P268v+IA/OtuOPQFfPU+wMDmt8K/bsFq2PKW7XtdWfv6VvKlc61VVpPqJaiA6AXUeBp4Z8MB/u+9bdz811Wc8asl/N/72/AbuG3WcF799mkMyzxGZrSR5ONH4bWbrT04GGNC/2gP5cODg2H3cru99gX72trg4vPCE7Phze9B0Wb4rCcLiCX29fC21s998Rp4+jz4VR68fpsd7MIt9vnRo2CctcDKC9rX11BUFYGnGg5+YQVDTAJseRuqD9u+vfYf8JuxsPZv8JXvwoy7IGM4bP5X+Pd4/xeB99VF7etn6S5AwO+FvZ82f57fD8/Ns5pWMJWHYPF9sPXd9t2/nehUswez+WAFz6/Yc2SNZICBaQncesYJfP0rw3TB+7ayb7V93bUMJlwe2P/FP2DBd+GmRTDgxMD+kh12dv3ez+GKv8D2xdZWv2e5tdsnNy4meITN/4LCfLj4d3bW+c58OLwDMk/ovGfZ/Ql89Ahc+nigH8bA9vdgx/tw1k8grhW/kzGwwxEQxa0IiPpKKNsN4y6F+BTY+AasexEmXAGXPwXBuTFffgifP2/bxKfAsBmw9nkYNhN2fxQQEPWVVoMZNQeiY9v+GdSWwaMTwOeBmERI7AvznoBn58KGf0ByNqx/CabeBNO/A1kjbLuxF8Pyx2z7xL4t3+PLZVaIjjwXtv3bCojs0W3va+mXkHeGNTHt+ghOODP0eXs/hR3v2e/Z+Mvsvk+fgMX32u9Sv0Uw6ry237+dqAbRw6hv8PHPtfu48o+fMOfRZfx91V6+Oq4fL94ynfz7z+OTH57N/PPHqHBoK95aqxEA7Pyg8bF1L0JDHfzzdjv7d6krt697lttjxg8X/sa+bnm7+Xt99pS1gU+5HsZcaPdtWdh5zwKw9R07YD1/pZ1Bf7kMnp4Dz18OKx6DdS+Ebuetg0X3Qtle6zSt3A/pudbsU1/Z/P2Kt9rXiVfC3MfgB5th5vftQLzi8cB5RVvghWtg27tWoGx/D968035mF/yvPafcWW5+1V/g79fDn84IaGlN8VQ336eSnfb/Nm4ujDgbrvyr9T8MmGR9RYvvg34T7H1d4QAw5mLwN9jPrzU+/aMVNLPn2+2m2mcojLGfZY2zhI3fbzWI/hNh4JSW/RD5/7SvhZvsq68BFv3MPtOpt8GhDdbR30WogOhBLPziADMeep/vvbSWwsp6fnzBWD794dk8cvVkvnJCZs9YGa2+Evat6e5eNGbti3ZAaIkD662JI76PneG61JTAlx/YH+7B9fDJ7wLHXAGRlGVndUOm21ld2tDmTRSH8u0seepNNiqo71A7MLTF5g3Q4LGO1OZMOKW7IS4FDqyF306Gv15kB44Lf2MHk8+eCt124+uOqe1WO3gDnHKLfT28vfn+FG2xr+7sOT4Vzr4XRl9oB7Dt79nP8uVvQGwifPsT+PbH8J9b4cZ34RsLoN84OzN2NYiiLfb/UV8JfzkfNrzW+J47P4CHhlofgEuwAHcFzcy74ernYNhX7Pbk65yBdDec+8DR0VmDTobUAfZ/XdWCyai+ErYtgvHz7P8crObYErs+hv/JtabJX+VZ01fVISvI0nMhd6bVZEMJPr8/ICCKt9pnPbzNarEn3xD4P215x74WbYn4b1EFRA/A5zf8zzub+c7zaxicnsSzN05jyQ9mc8sZw3veymiv3wZPngWHNratXcUBeO8XLc9Sm7L+ZdjTgr3WZemDsOw3LZ+z3/khTb3RqvvuLGzLQjubvPA3dia69KHAIFBnw4Y5y3F6TrnOmlLGXGhNM/VVR99n1Z8hOh6mBEUEjbnIOncP5dtomDXPtf5Mr90Cvx4NDw2zn7m/yUp7ZbthyDRrxopPgfMehDs/h1NusgNJ0SZrhmrKmr9CbBLs+QSW/Le1x4842x4rbkVARMVCel5gnwjM/b0dbP82zw6IRVusyanPQHtOVDQMnQ65M+x22uCAgDi8HfqfCN9ZYc957VZrHnP58H/t/2bDq07/ttmBd9fHzmfgCIi+Qxr3deKV9n8w8tzQppyoKLjw1/Z5nzwroFk2Zeu7Nupp/GWQlAES1bIPwtcAb91thd5ZP7H7tr9nv28AGXmQe3rzfoh9q6xGN+Kr9pzDO+zEBuznlDUCMkfA1rft5OWZC+HJM+EvF1rzZwQWf1MB0Y34/YY31+3n3Ec+4PGlO7h22lBe/o+vcMao7JDlsbudPSucmbOxg30wfn9gltmUQxvhqa/CsodbNs0Es+E1O0g+fa6NQGlOsFQetINl6a6Wo0P2rbYD2YlX221Xi8j/p53lD5wCk6619mxXeNSVQ1wqnHQDXPsSTHKWVB97kR04mpoo6qtspM6EeZCcGdg/5kLAwJ9Ot7b4jx45un+rnrYzcWOsuSX/DesnGXWeNYGteabx+aW7oe8wG5p65+fwle9ArF1rgwmXQ0Jf+OwJa+N/98f2cyraYs1ls/4fjDofPJUw/EwrJCQqYEYKRfFW60NpGp6alAG3vAfznoRZ8+GaF5q3rwOkDQkSENvsNeNT7OebPRpeut46vwtWW19RTAJsftN+LmuetTPxAmcl4vK99v+T0MSPkJQBNy+2/ojmGHMhfGuh/T++enPz2lbqABhyqhV0SZlQ3YKJadXTNihhzoNwxj2QNdqak9wQ1/S8QB5FqAlW/j8hOs6a7sD6sQ6ss59BlpPYN2qONSf++6d2IjPz+9bUtuTB5vvVAXqAzeL4pKiynrtfXsuybcWM6pfCH68/mTkT2hCX3dUYYwewlP7Wtr7sYTu7H3qqPb76LzbO/QdbICUn0K50N/z5vEDeQDj209JdNgJo8CkweBp8+rh10F32x6PP3bPC7SAUb7EDPVgVfsEd9n43vGVV8UEnQ85Ya1Pe+YGd2e9YAtNvs7PhxHTb1g1lrK+AhDQ74xx9fuCeQ79iB44vXrHCwGXj6+CpgpO/1biP/SZAv4kgWDPVyidt2GNG0Gx87Qt24IuKtYN66kC45PfWXFN5wNrTx1xkP9v6SqgtgfRhoT+/uCT7P1r+eytowM7MB50EUTEw+WtWGL5SBpOugZh4KyRbimQq2gL9J4Q+lpJjc0PCIc3RAGpKoOYwZI20+xP7wvWv2iip56+A7DH2s5/1/+DdH1lT2vq/23NLdtjXsr32eqGKRwYHGzTHoJNg1n/BWz+wJqngBLa6CmtemvqtQN5DcnbzJqaaEljyS8ibFfA7DTvNBkAMONEK4LQhEBNnBY0bFOD3Wc2p6pDVFk44CwZNtecXbrJmz37jA4J59Pn2/7rmr3DSN2zY7uwfQdXB0J9DB4moBiEic0Rki4hsF5H5IY4PFZElIvK5iKwXkQuc/bkiUisia52/ECND72RXcTVPLdvJ+b9dxspdJfzysgm8/b0zeo5wqCkJHYa4ZaFVi2fPh9PvhuQceO/+wMxrw2vWEVlzuHG7TW/ameoNb1lbvms3bg5j7A8Ggcv/DHP+25qFNrwGtaVHnx+sqhdutq/l++xAs+E1O+guutcOKgOn2B9R3hnWJ/DkmVaVH3epbecKCPc+deV2kGpKVLQ1Y2z7d+MB4/Pn7ExvyLTG54vArUvgP5ZZRyNYn0Yw5QV2prjsYWtqOOvHdqAXgYsesU52N76/dLd97duMgAAb0jnhCrj0j/C1V6w55/O/2QEmJQdS+8GN7wT6mjWqeROTt86aSbLaEb3TlLTBUF8eyLvIHBk4ltofvv66FZJ7lltT2cSrAIGF99hBNDreml4AyvccbV5qK+Pn2fute6nx/q3vBMxLLsnZzTupV/7ZTizmPBgYqHNn2u9+/gLoM9gKB/eZXX9P8Vbr6K8uthnm0x1NMOMER4NYb81LLkOmW40pIc36gMBet2liYCcRMQEhItHAY8D5wDjgWhEZ1+S0nwAvG2OmANcAfwg6tsMYM9n5uy1S/exK/vfdzcx+eCkPvLWJQX0T+OftM7nu1GFEd4Y5ac8K+0PuCH4/PHcZ/O3yo4+tedZ+yad83WoDZ/yndcbuWWF/NLsdu7C3iZln5xI7+GSNsF/i1jSIkp120D/zh4EZ8knfsD/W9a9YO+9z8+Dt+YHnHjzNquaFji35nfl2hn7dK3Yg/9SJshl0snO9b9pZWc44mP3DwH7XVNFIQPQJ3c9J1za2jxdttf2e8vXQM7noWLs/8wQ7sG8PEhA+rzUBfeV2qyUM/Yq9vkvWSDtYbviHNWOVOQKiOQ0CoM8AuOLPMPlaGHWujfBJyoLpt4c+3x20mvo6wApX429feGdT0gbbVzf/InNE4+MZw62QmPQ1+3mkZFv/RMFKO/Mef2lgcC3ba2flHSEpw/oqvviHnc2DnaSsetp+3wcHCfvk7NA+CL/f5lnknm6/Vy7DHL/L4W2QkRvYnzUioEG4pqYr/mzLfwyfZbdzxlott77cBh24RMdYv8+VzzQfZt2JRFKDmAZsN8bsNMZ4gJeAuU3OMYD7C0wD9kewP93Kvzce5LElO5g3ZRDL/utM/vndmYzun9o5F1/5ZztjXnxfx66z7gWryhdttolGLp5qa/YYe1FA1Z3ydVv07JP/C/gloLEfoKHemhOGOzbpcATEzqX2deS5gX0DJgVCFz96xM6+P/uTnV0dXG8doFmjbL99DfYaEy6HkedY56078Lvmp+Gz4OZFcM3zViNyB/TEUAKimTLn/cbZmd26F+32589Z882ka1p+PhHrhNz5gY1UAqjYDxgrOK55Hr719tGRNyecaQfpA2uDNIjclu8VzJgL4J7tgUifpmSNsNEyFSG0x6YRTB3BHdB3LrWfVygh138CXPa4HbzBCk2wwj57jNUkKg/aGXtHNQiASVdbE4373dv+ntVgTv9+47IaKTmhBcRux89w0jca7+8zwAo8sBFMLpkjrS+jrtyatqJiG2tSYCcvHsfv1tRcNvZia4rqAiIpIAYBwfaEAmdfMPcB14tIAbAQuCPoWJ5jevpARE6PYD8jzvbCKn7wyjomDkrjwcsndm7hvO2LrfodHW9NCG7kTVupr7Imo+Rsu13wWeDYjiXWORhsh49LglNutqanFX+0P3awphCXvZ/aQcd1WvZ1HJTBDsHKg/DMRdbxBtZ53GdQ4IflctI3bKmEpf8NI8+zz/vqzXYWP2S6nXEVbrLO6PqKwD1Tsq3v4rQ7W0+KiomH2GSbQAX2BxzfjAYBVhjs/xxe/zas/qt1IAb7X5pjxFfBWw17Hf9JxT776s6uQ2kgAx3nZsEqq0HEpQQG0HBpyUbtOkFDJcwVbwXk6Nl+e3Cf8eAXdtAMJ0FuwuUw9DSrRbl9cAfzjmoQYL9P8Wk2n8NTDe/f7+SxNBnwk7Osj8lTY81ci35mTbJrnrPtx1589LVdLSI4+sv1uxRvtxpE9uiA+cklZ6x9lWjIGU930d1RTNcCzxhjBrz5K3oAACAASURBVAMXAM+JSBRwABjqmJ7uBl4QkaN+qSJyq4isEpFVRUXtTIGPIBv3l3Pbc6s595EPEOAP151EfEwnVsysr4JXbrSzjetesTMOtxxEOG3dGSzABw/ZmdkVT9vBfm+QgNjytv0BuF92l2m3WtNO8ZbAjN8bFN+9Y4m9Vu5Mu913mBU0rh3X1wD/uNFGq3zwP1ZV//JD6+hrOphNuMJmyyY7A/7UG+19wdrRs8dY/8amBYDYa7iMPh/ObRJ11RyJ6Y0FRHMaBNgZbXyaFZL9J8CZP27+3GDyTrefi2tmKm8iIEKRnGkHmX2rAhFMnemUdGewoSLRijbbmX5sJyRnpvQLTCaazpqbo88AuPFtJ8zTyUbvTAERmwAzvwfbF9nChQfWWdNj00E72RH+1UW2ztbHv4U/zrTfuROvDP35uL+Z4IAE97kPb4ODGxqbpVxyHGt89phAdFo3EEkBsQ8I/u8NdvYFcxPwMoAxZjmQAGQZY+qNMYed/auBHcBRBdyNMU8YY6YaY6ZmZ2dH4BHah9fn57eLtzH39x+z4svD3HrGCbxz1xmdX3K7Yp+1Uc74njWbDD7Fml5C2ZGb8uI18PuTbQz4qqetqejkG6wDt//EQCih32cddiPPOXq2l5JtbdwQCB8N1iB2LrF9indMaa4jzTUzvf8L67sYepoVEhtfs9E5eWcc3d/EvnDVX+Frf7cz5xl3Wi0ia5Tddn9Qa56z5ZzbOrsOvk9tqZMNW9GygEjJgf/aCf9vlw2Z7NfUxdYM8ak2UmWPkz3sOu77NFWwmzB4qo3GKtvdsv+hPaTkWK1tW4haP0VbO8dBDdZ05uZItKf0iKtZuiVCOsPEBHD6D2wgRXK2NR263+dgXO26utgKkb5D7W+ioc5GjYVizIV2IuWaWcFqThLtVJXdH1pAZAy3QQvB/oduIJJhriuBkSKShxUM1wBfa3LOHuBs4BkRGYsVEEUikg2UGGN8IjIcGAnsjGBfOw2f3/Dtv61m8aZC5k4eyP2XTCAtqR11ZsLBtZW7g+Gpt8GrN9mZUEv1Who81rnr99r8BG+NVbMveNgeHzzN2tR9DVZQ1BRbG3Yozr7X1tgZ6ti23QzRmhLYv9bOxFzc2V75HitcPn7Umo7O+pktqLbwHns8lICAxs+U2h8u+b9AvSF3jYH68o7ZZxPT7efqqbI2/+ac1C7tLVvdf6IN2zTGCvqEvjYfoCUGnWxDa6sONf8ZtRcRG9Hz0W9sdnGKMxj6fdYp7CbTdQZpQ+wkIStMDSKY2EQnl2Kvtd2ndGL0X+5Mm7Tn94VeG+OIgCi0vq+xF9tM7aKg8OqmJPQJlBhxiYmzQmLTArsdSkBEx9icks4w63WAiGkQxpgG4LvAu8AmbLTSRhG5X0QucU77AXCLiKwDXgRuMMYY4AxgvYisBf4B3GaMKYlUXzuTX727mcWbCvnpReP47TVTIiccIFDrxRUQ4+baSI8vWilpXbTZCodzH7Cz3rwzbFSEqyEMPsUKjcKN1gkbFWvt5qFIyrDqtTtQuxrEvjWACWTQQmC2V7YnkKg23YlUGXeJ1R4yR0BaKzNpl0lXB+y+fXOtCQoaz9baiqtBuL6cljSIjtBvnNVQyvdav0xL5iWXQVPtq7+h5RDX9jJhnhWKbu4EWOerr75zHNQu7rO2d/BztYi0Qe1bm6ElRJoX+q7Q3P+5/Y4MmGS/H03DmsMha2TA4d2vmQWERpzd+ZpiG4loopwxZiHW+Ry872dB7/OBGSHavQq8Gsm+dTbGGP726R7+9MFOrjt1KDfNzGu9UVPqKmxE0KRrw7Mv1zoCwo3fj461jtLN/7Khk805AA9+YV9Hnmfj5aHx/YacYl8/+JW91rT/aH2gjHUFhBPF5CabJQeZ/uJTbeRT2R5r2krOCQw8U2+yIaPtnRlHRVktomhL+36wLonptu9uHaaICQgn6ezQRuuDCEdA9J9ohbXfG5mBI2ectXlvfB2mOXV/XJ9EZ5mYoOMCInOErZ/VGf6HtuB+l7cvtq/9O2D+yRppTbdJWeEFNnQT3e2kPiY4WF7HTX9dxU/f2MDpI7O475J2Rh188D/wxrdbrhcfjGtiSgyyt4+50A5ubl6CS4MnEOd9cL0d0DNPsIKhqTDqO8wO3pv/ZQeNc37eel+ioq1PwBUQHqdOUVwTs0lfx7ywa5lV6d17DzsNvvrz5uP0w+GUm22Jg5j49l8jwdUgHAHRUhRTR3CjVA5ttFpEa/4HsM5KN5s5EhqEa2ba/YkTeksgEKAz13Cecj3Mecg6rNuDK1i6WkDEJtrSHvvW2EznUKahcHEd1f3GRyQDurNQAdFBPA1+rntqBct3HOanF43jmW9NIza6HR9rTYktfwwBB1w4baJiAk5gsOaVmMSjq4c+dTa84/gDDn5hv5jNrUEsYpOTouNtNnO40StxSYE8CNcX0XRpzr5DbYmOygM2mif4njPvalyWua1Mud5meXeExHQn0uqQ3W5a56eziE+1n0XBSquxhKNBQMDMFKHMWVs6xASqihZttWVFOlOTyhgO07/d/oHRdW53loO6LSRnAcYGR7S23kZLuP6Xfs2UL+khqIDoIE99tJMdRdU8dt0UbpqZ1/6s6M+esCGiqQMDIXytUVtiB7TgH1pcks0B2LwwkG/g89qEnHUvWh/BwS8ap++HYs5DNpks3MgcsFqJ64Oob06DGBZIAMrtgektrrnOzVZuzUndEfpNCKxNEa6AmHEnXPanyPUra6Sd3bohuEWbAzkSPYWcsXYG3x39cs1Brf1+WiNnnPUXDp/d0R5FFBUQHaCgtIbfvbeN88b346wx7VSXwQ6mn/7RVticfK0zqwwj4a22tLF5yWX0BTYj9qBTKrhin3U+1lfAp3+yr60VM0sb1PYQu9ikQB6Ep8qG6TV1+LlmgZT+3R6hERJXQLjZypHyQYAdJBocgRqOiQms5tBatnZHGT7LmpkaPDZxLntMZO/XVvoOhW8vb1wnqatw/RAdDT9N7GtDpEed2/q53YgKiA7wy7c2IQg/u7gVW2Txdlu0rjny/2kH+5l3WROR8TX2Ifga4LMnj661VFMSOt5/9PmAwFanHHVweQt33YT+zUROdITYxIAG4ak6WnuAgGkk7/SeaXt1s63dEs2R8kFAYxt2uBpEV5A3ywr6zU6hxc70P3QWOWOaN5FGErf+UTjVYo8BVEC0k/z9Fby94SC3njG89eU/P34UXrmhcRJZMAfW2cF08DQbgROb1NgPsWuZLaXtljt2qS0NzHiDSc6ycdaFTiEwdzY89mKbJyDRgcSyziQuOeB78FQf7X+AgP04ONO5J3FEg9hlfTCRzGI9IiAkkDzWE8idCYidlEDnRjD1dvoMsuatSEyweiAqINrJ75dsIzU+hhvDCWct3WVj190Sx00pzLdqfFSUjcAZdlqg2iUEMm2bLlDTnIkJnPLNTl2dsj32Sz3j+4FjnVE2oSnBGkR9VWPnuUv2aLhhYeNqpT0JV0CU742seQlsSefoeJv0F05Noq4iKcPOkN1M787MgejtnHIzfPPN0BOzYxAVEO1g66FKFn5xkG+elktaYhg/bNdcEVzfyMUYG+oY7AwePtsWSHNDDd31GXYsaWxmqilpvgBdVlD55rI9duYz6CSrpbS04ldHiE1qHOYaSoMAmzzX3gzkSOP+8H2eyAuI6BjrcO3qcM1wcPNREtMb57Ic7yRlBGqLHQeogGgHjy3ZTlJcdHjJcA2eQLVOV0BUHICNTrZqVaGNRgo2+bjrE7jrG7gCwlttSwuDnak31DZfcyhrlA3XLN9rI3L6DrU2/xvftYuaRIKjBEQrpSN6InEp1gQHkY1gcrnkd3Z95J5G3mz7mjW6Z/qKlC5BBUQbKayo4631B/jatKGkJ8e13qB8r40gik2yJbSNgUU/hVe+aWf2rhAIFhBuKYESZ7Hzsj02rC4m0S6kDqGT5II5UlJ4m23vJlZ1dmmCYJrmQTSnQfRkgpcejbQGAc5aFz3Q4Tl0us3azulhEUxKl6ICoo28tHIvDX7DddPDzGR1zUujL7C1V/atCSQhbVsUEBDBES0p/axAKXHqE5YX2AF/+Gybnm9MoA5Tc7ZQN0a8cKM1VUUqsSqYpnkQvVGDgMBnGskIpp5OfApc/w+bma4ct6iAaAMNPj8vfraH00dmkZcV5uzYTbhyywe/dXfAvu0KiOTsxssHitja/yVfWh9ChVOrZ9S5Vhso2hyow9SciSkp0w50O5ZgVyvrKgFRbQWYp6r16qQ9la7UIHoyw2f3rPBbpctRAdEG3t9cyIHyOq47tQ11cEp32UV1TjjT1nE5sNaugDbxKltwbP/a0CGnGXlWg6gusgIlbUgg63Lvp62bmESsFuHmU3SJgEi05rSG+pad1D0d1/F/vAsI5bhHBUQbeP7TPfTvk8BXx7ah+mLprsDCIoOcZSOnfsuuwOatsSUwmlswpPTLgAaSNsSWtI5NtpVQWzMxgS2Z4HNWjesKAeEKhLoyG9bb201MXeGkVpQeTA+NNex5HCyv48NtRdxx5ghi2lKMz10eEuyCNyU77boNxthSFA11gcqewWTk2cHdjXxKG+yUtB5rzVKpTmmPllZOcx3VEh1+KYeO4OZWuEuK9noBEaFCfYrSS1ANIkzeWLsPY2DeSW20yZbuslnNAF+5Hb633ikbnBQoVhdqUXI3ksldWMe1BeeMtXkTNSU2qqmlhDfXUZ02qGvyDmIdDaLaERC93QdxPDupFQUVEGFhjOHV1QWcPCyd3Jac03UVAd8AQG2ZNbe4AgIah5meeJVdMCRUKKErIHZ/Yn0Xrj2833jroC7a0no2pysgIrF2QCiOaBDOSlm91QeRoD4IRQEVEGGxcX8F2wqrmHdSK2aaBXfAy98IbLv+g+ZW/zrxKrhne+iBtM8g69z2VNq6926ykuvQ3rOiZfOSe9+o2K7L1HXr47saRFyIUhu9AY1iUhRAfRBh8eqaAuKio7hoYpOCau//0ppRZnzPbu9b3djk4+ZABGsQTWkuSzUq2rYr3to41NB1aHsqW9cgomNtpm5XLUriLjta3cs1iEEnWbNfT1sHQVG6GNUgWsHvN7y57gBnj80hLSmo7pIxsOrPsOZZu11fabOmgyu2ulVUWxIQLZHulPIIFhDJWXY5UAivYNjkr3Vdpq4rIFwTU2/1QWSNhO98AsmZ3d0TRelWIiogRGSOiGwRke0iMj/E8aEiskREPheR9SJyQdCxHzrttojIeZHsZ0usLSijuKqeORP6Nz5QsQ9qDsPhHTZr2F3c3a1FBNbElNC3/aYK1w/RNFnJjXpqzcTU1RwREM5ynb1Vg1AUBYiggBCRaOAx4HxgHHCtiDTNCPsJ8LIxZgpwDfAHp+04Z3s8MAf4g3O9Lue9TYeIjhJmj2qS+3DAWa0Npxpr4Sa76QkSELVlHSsLfERANMlhcM1MzSXJdRdxTU1MvdQHoSgKEFkNYhqw3Riz0xjjAV4C5jY5xwBuLGEa4NS3Zi7wkjGm3hjzJbDduV6X896mQqYOS29sXoLAcp7u+6LN9n1DrS2PAVab6Mgs2hUEbj6Di+uo7mk16Y/Kg1ANQlF6M5EUEIOAvUHbBc6+YO4DrheRAmAhcEcb2iIit4rIKhFZVVRU1Fn9Dty0tIbNByv56tgQ600fWG/XVE7MsALCLboHNvkNbEVT1+zSHnJnwO2fwcDJjfe7giO4flNPwM2DqCm2CxRFYlEiRVG6jFYFhIhcLCKREiTXAs8YYwYDFwDPteVexpgnjDFTjTFTs7M7f1GT9zfbmfDZoUprHFwfKNV8YD0Ubg4ccx3V3tqA2aW9hFrNa+AUuPh3dgnRnkR0rM3aNn5nXQVdR0BRejPhDMZXA9tE5Fci0pbi8PuA4AD8wc6+YG4CXgYwxiwHEoCsMNtGnPc2FZKXlczw7CbRODUlNmKp/4l2bdpDG6Fyv9UowFY0BWtiio2AmUUETv5m6CU9uxORgFmpt5bZUBTlCK0KCGPM9cAUYAfwjIgsd0w7rY1OK4GRIpInInFYp/OCJufsAc4GEJGxWAFR5Jx3jYjEi0geMBIIsV5n5PD5DSt2HmbWqBCaiet/GHAi9J8Efq/dHugU43M1CE91xzWI3oZrVlL/g6L0esIy5xhjKoB/YB3NA4DLgDUickcLbRqA7wLvApuw0UobReR+EbnEOe0HwC0isg54EbjBWDZiNYt84B3gdmOMr11P2E72lNRQ3+Bn3MAQ9XgOrLOv/SdZDcLFXSrUDXX11nTMB9EbcZ+3t+ZAKIpyhFYzqZ3B/FvACOBZYJoxplBEkrAD+P8119YYsxDrfA7e97Og9/nAjGba/hL4ZRjPEBG2F1YBMDInxEB3YD30GWwTqRL72qJ5IpDtZN4e0SA6GMXUG3EFhJqYFKXXE06pjcuBR4wxHwbvNMbUiMhNkelW97OtsJL+HOaE9CbpFz6vLaDnru0QFW1NTa5jFqxgMMb6Io43DSJOBYSiHCuEIyDuAw64GyKSCPQzxuwyxrwXqY51N9sLq3gr4af0+XgTzHkwcGDjG9YhPeU3gX2XPm5fg01LDfWO0DjOBIT6IBTlmCEcH8QrgD9o2+fsO6b58lAZmZRB/gKrDYB9/fi3kDUaRgZV/8g8wf652oK3NiAsIhHF1JNxn1d9EIrS6wlHQMQ4mdAAOO/jItel7sfvNxx0E+8qCuyyoAA73odDX8CMOxuv6+ByREDU2AgmOI41CBUQitLbCUdAFAVFHSEic4HiyHWp+zlQUUeMtzKwY8vbVnv46BFI6Q8Trwzd0B0cvTVBGsRxJiDUB6Eoxwzh+CBuA54Xkd8Dgi2B8Y2Wm/Ruth2qJBW3bLdYAZE1EnYtgzn/AzHxoRsGaxDHq4A4EsV0nJnWFOUYpFUBYYzZAUwXkRRnuyrivepmthdWBQRE3ul2XeiF99jM6VNubr6hW2rCWxuo6nrcmZg0D0JRjhXCWlFORC7Elt5OEKe+jjHm/gj2q1vZXljFgESPdc2feLUVENXF8LW/Q3QLH5mIHSCPaye1mpgU5VghnES5PwJJwJnAU8AVdHHZi65mW2EVs1N9UA4M/Yotjpd7eiBTuiXikqyD+nh1UqsPQlGOGcLRIE4zxpwoIuuNMT8XkV8Db0e6Y92FMYbthVV8c6AjIOL7wK1Lw79AbGITDeI4ExCaB6EoxwzhRDE5ixtQIyIDAS+2HtMxSY3HR3mtl35xTmRvQohaTC0Rm9QkzPU4Gyg1D0JRjhnC0SDeFJG+wP8Ca7CrwD0Z0V51IyXVVjCkSS1ExzcfsdQcx7sGkTMGknOg77Du7omiKB2kRQHhLN7znjGmDHhVRP4FJBhjyrukd93AYUdApFDTvvUWXCe15zgVEAOnwD3bursXiqJ0Ai2amIwxfuCxoO36Y1k4AJRU1wOQRE3bzUvgCIhq+xeTGDrjWlEUpRcQzuj1nohcLnJ8rB95uMpqEAm+auugbiuuiclTc/xFMCmKckwRjoD4D2xxvnoRqRCRShGpiHC/ug3XBxHnq+qYiclbe/zlQCiKckwRTiZ1D1v4OLKUVHuIi44i2lMJKcPbfoE4J4rJWx0I+VQURemFhJMod0ao/U0XEDpWOFztISM5DqmvbL+JyVOjJiZFUXo94YS53hP0PgGYBqwGzopIj7qZkmoPmSlxUF3Rfid1Q63Ng1ATk6IovZhwTEwXB2+LyBDg0XAuLiJzgN8C0cBTxpiHmhx/BFvCA2w5jxxjTF/nmA/4wjm2xxhzCV3A4WoPmUkxUFLRTh+EY1aqOQzpmgugKErvJaxifU0oAMa2dpKIRGNDZM9x2qwUkQXGmHz3HGPM94POvwOYEnSJWmPM5Hb0r334/fDqTQyqmErq0MmAaaeJydEaaoohp9WPSVEUpccSjg/i/7DZ02CjniZjM6pbYxqw3Riz07nOS8BcIL+Z868F7g3jupGhvhw2vsZU46U8Ybzd1y4Tk6tBlBx/ZTYURTmmCEeDWBX0vgF40RjzcRjtBmEXF3IpAE4NdaKIDAPygPeDdieIyCrnng8ZY94I457tx2vXf+jnP0RivE2Wa7eTGgBz/GVRK4pyTBGOgPgHUGeM8YE1HYlIkjGmphP7cQ3wD/ceDsOMMftEZDjwvoh84SxedAQRuRW4FWDo0KEd64EjIAZLMYWxHREQQUJBo5gURenFhJVJDQQH9CcCi8Notw8YErQ92NkXimuAF4N3GGP2Oa87gaU09k+45zxhjJlqjJmanZ0dRpdawBEQQ6SQjBingG17TEzBQkGjmBRF6cWEIyASgpcZdd6HMzVeCYwUkTwRicMKgQVNTxKRMUA6sDxoX7qIxDvvs4AZNO+76BwcAZEhVWT6iu0+1SAURTmOCUdAVIvISe6GiJwM7oLNzWOMaQC+C7wLbAJeNsZsFJH7RSQ4ZPUa4CVjjAnaNxZYJSLrgCVYH0SEBUTAYpZeudW+6UiYK6gPQlGUXk04Poi7gFdEZD8gQH/g6nAuboxZCCxssu9nTbbvC9HuE2BiOPfoNLwBmZdcusm+aW+inItGMSmK0osJJ1FupWMGGu3s2mKM8Ua2W91AkAYRXZQPEtW+dZUbaRBai0lRlN5LqyYmEbkdSDbGbDDGbABSROQ7ke9aFxOkQUi9k0XdngrnseqkVhTl2CAcH8QtzopyABhjSoFbItelbqLBCogyccxK7XFQgzqpFUU5ZghHQEQHLxbklNCIi1yXuglHgzgQ69RPaq+AiI4Fibbv1UmtKEovJhwB8Q7wdxE5W0TOxuYrvB3ZbnUDjoAoTsyz2+1xUIM1S7mCQZ3UiqL0YsKJYvp/2Gzl25zt9dhIpmMLbw0eYqhJGQrltC/E1SUuCTyVqkEoitKraVWDMMb4gU+BXdgCfGdh8xqOKXz1NdSaOLypTsmO9pqYIBC9pD4IRVF6Mc1qECIyClth9VqgGPg7gDHmzOba9GZ89dXUEo83dbDd0V4TEwQ0B41iUhSlF9OSiWkzsAy4yBizHUBEvt/C+b0av9dqEPWpTvmojmoQUTEQc+z58hVFOX5oycQ0DzgALBGRJx0HdTsSA3oHxlNLHXGQ0BfO+C8Yf2n7LxabpNqDoii9nmY1CGf9hTdEJBm70M9dQI6IPA68boz5dxf1sWvw1lJHPLHRUXDWjzt2rdgk9T8oitLrCcdJXW2MecFZm3ow8Dk2sunYwltLrYkjLiacyN9WSOgDCWkdv46iKEo30qY1qZ0s6iecv2MLbw21rgbRUc78EdSWtX6eoihKD6ZNAuKYpqGWWjJJ7gwNImN4x6+hKIrSzXTCaHhsEBXsg1AURVFUQLhIQyf6IBRFUY4BdDR0iPLVUUccsdHHbCSvoihKm1ABAWAM0Q211KIahKIoiouOhgA+D4KhzsQTpz4IRVEUQAWExVluVDUIRVGUABEdDUVkjohsEZHtIjI/xPFHRGSt87dVRMqCjn1TRLY5f9+MZD/dtSA6LQ9CURTlGCBieRDOynOPAecABcBKEVlgjMl3zzHGfD/o/DuAKc77DOBeYCpggNVO29KIdNYVEBrFpCiKcoRIjobTgO3GmJ3GGA/wEramU3Nci12tDuA8YJExpsQRCouAORHrqWNi0jwIRVGUAJEcDQcBe4O2C5x9RyEiw4A84P22tBWRW0VklYisKioqan9PHQ2ijjjiVYNQFEUBeo6T+hrgH8YYX1saGWOeMMZMNcZMzc7Obv/dXSe1iVMNQlEUxSGSo+E+YEjQ9mBnXyiuIWBeamvbjuNoEJ6oeKKjNFFOURQFIisgVgIjRSRPROKwQmBB05NEZAyQDiwP2v0ucK6IpItIOnCusy8yOAKiISoxYrdQFEXpbUQsiskY0yAi38UO7NHA08aYjSJyP7DKGOMKi2uAl4wxJqhtiYj8AitkAO43xpREqq9HBER0fMRuoSiK0tuIaLlvY8xCYGGTfT9rsn1fM22fBp6OWOeCcXwQvmhdBU5RFMVFPbJwRIPwqwahKIpyBBUQcERAmJiEbu6IoihKz0EFBIC3Bi+xRMfEdndPFEVRegwqIAC8tXii4rXMhqIoShA6IgI01FIvWmZDURQlGB0RAby11KMahKIoSjA6IoIVEKKLBSmKogSjIyKAt4Y6XSxIURSlEToiAnhrqTPxxEZrHSZFURQXFRAA3hpnudHo7u6JoihKj0EFBIC3lhoTpz4IRVGUIHREBPDWOsuNqolJURTFRQUEgLeWatUgFEVRGqEjIhwxMWminKIoSgAdEY0Bbw1Vfg1zVRRFCUZHxIZ6wFDrj1UNQlEUJQgdEZ3Fgmo1UU5RFKUROiJGReOZ9HU2m6HqpFYURQlCR8SENCrP+TXL/eNVg1AURQkioiOiiMwRkS0isl1E5jdzzlUiki8iG0XkhaD9PhFZ6/wtiGQ/vT4DoD4IRVGUIGIidWERiQYeA84BCoCVIrLAGJMfdM5I4IfADGNMqYjkBF2i1hgzOVL9C8bT4AdQDUJRFCWISI6I04DtxpidxhgP8BIwt8k5twCPGWNKAYwxhRHsT7N4fFZAaLE+RVGUAJEUEIOAvUHbBc6+YEYBo0TkYxFZISJzgo4liMgqZ/+lEeznEQ0iXjUIRVGUI0TMxNSG+48EZgODgQ9FZKIxpgwYZozZJyLDgfdF5AtjzI7gxiJyK3ArwNChQ9vdCe8RDUIFhKIoikskR8R9wJCg7cHOvmAKgAXGGK8x5ktgK1ZgYIzZ57zuBJYCU5rewBjzhDFmqjFmanZ2drs76pqY1AehKIoSIJIj4kpgpIjkiUgccA3QNBrpDaz2gIhkYU1OO0UkXUTig/bPAPKJEN4G1SAURVGaEjETkzGmQUS+C7wLRANPG2M2isj9wCpjzALn2Lkikg/4gHuMMYdF5DTgTyLixwqxh4KjnzqbetUgFEVRjiKiPghjzEJgYZN9Pwt6b4C7nb/gn665LQAAEi9JREFUcz4BJkayb8G4GoRmUiuKogTobid1j0B9EIrSdrxeLwUFBdTV1XV3V5QwSEhIYPDgwcTGxobdRgUEgSgm1SAUJXwKCgpITU0lNzcXEc0h6skYYzh8+DAFBQXk5eWF3U5HRAJ5ELGqQShK2NTV1ZGZmanCoRcgImRmZrZZ29MRkaBSG6pBKEqbUOHQe2jP/0pHRMDjFOtTAaEoihJAR0S0WJ+i9EbKysr4wx/+0OZ2F1xwAWVlZRHo0bGHjogEl9pQdVlRegvNCYiGhoYW2y1cuJC+fftGqlsdprX+dyUaxYTVIKIEYtTEpCjt4udvbiR/f0WnXnPcwD7ce/H4Zo/Pnz+fHTt2MHnyZGJjY0lISCA9PZ3NmzezdetWLr30Uvbu3UtdXR3f+973uPXWWwHIzc1l1apVVFVVcf755zNz5kw++eQTBg0axD//+U8SExND3u/JJ5/kiSeewOPxMGLECJ577jmSkpI4dOgQt912Gzt37gTg8ccf57TTTuPZZ5/l4YcfRkQ48cQTee6557jhhhu46KKLuOKKKwBISUmhqqqKpUuX8tOf/jSs/r/zzjv86Ec/wufzkZWVxaJFixg9ejSffPIJ2dnZ+P1+Ro0axfLly+lICSJQAQFYDULLbChK7+Khhx5iw4YNrF27lqVLl3LhhReyYcOGI2GcTz/9NBkZGdTW1nLKKadw+eWXk5mZ2ega27Zt48UXX+TJJ5/kqquu4tVXX+X6668Peb958+Zxyy23APCTn/yEP//5z9xxxx3ceeedzJo1i9dffx2fz0dVVRUbN27kgQce4JNPPiErK4uSkpJWn2fNmjWt9t/v93PLLbfw4YcfkpeXR0lJCVFRUVx//fU8//zz3HXXXSxevJhJkyZ1WDiACggA6hv86n9QlA7Q0ky/q5g2bVqjGP/f/e53vP766wDs3buXbdu2HSUg8vLymDzZrkt28skns2vXrmavv2HDBn7yk59QVlZGVVUV5513HgDvv/8+zz77LADR0dGkpaXx7LPPcuWVV5KVlQVARkZGp/S/qKiIM84448h57nVvvPFG5s6dy1133cXTTz/Nt771rVbvFw4qILAahEYwKUrvJjk5+cj7pUuXsnjxYpYvX05SUhKzZ88OmQMQHx9/5H10dDS1tbXNXv+GG27gjTfeYNKkSTzzzDMsXbq0zX2MiYnB77c+T7/fj8fj6VD/XYYMGUK/fv14//33+eyzz3j++efb3LdQ6KiI9UGoBqEovYvU1FQqKytDHisvLyc9PZ2kpCQ2b97MihUrOny/yspKBgwYgNfrbTQAn3322Tz++OMA+Hw+ysvLOeuss3jllVc4fPgwwBETU25uLqtXrwZgwYIFeL3eNvV/+vTpfPjhh3z55ZeNrgtw8803c/3113PllVcSHR3d4ecFFRCA+iAUpTeSmZnJjBkzmDBhAvfcc0+jY3PmzKGhoYGxY8cyf/58pk+f3uH7/eIXv+DUU09lxowZjBkz5sj+3/72tyxZsoSJEydy8sknk5+fz/jx4/nxj3/MrFmzmDRpEnffbeuR3nLLLXzwwQdMmjSJ5cuXN9Iawul/dnY2TzzxBPPmzWPSpElcffXVR9pccsklVFVVdZp5CUBsQdXez9SpU82qVava1fY7z69m66EqFt89q5N7pSjHLps2bWLs2LHd3Q3FYdWqVXz/+99n2bJlzZ4T6n8mIquNMVNDna8+CMDTYNQHoShKr+Whhx7i8ccf7zTfg4uOithy31qoT1EUgNtvv53Jkyc3+vvLX/7S3d1qkfnz57N7925mzpzZqddVDQK7YFC8ahCKogCPPfZYd3ehx6CjIq4GoWU2FEVRglEBgeZBKIqihCKio6KIzBGRLSKyXUTmN3POVSKSLyIbReSFoP3fFJFtzt83I9lPT4OGuSqKojQlYj4IEYkGHgPOAQqAlSKywBiTH3TOSOCHwAxjTKmI5Dj7M4B7gamAAVY7bUsj0VdNlFMURTmaSI6K04DtxpidxhgP8BIwt8k5twCPuQO/MabQ2X8esMgYU+IcWwTMiVRHPWpiUpRjnpSUlO7uQq8jkqPiIGBv0HaBsy+YUcAoEflYRFaIyJw2tO00VINQFKWr6EnrPbRGd4e5xgAjgdnAYOBDEZkYbmMRuRW4FWDo0KHt7oSW2lCUDvL2fDj4Redes/9EOP+hZg/Pnz+fIUOGcPvttwNw3333ERMTw5IlSygtLcXr9fLAAw8wd25Tw8XRVFVVMXfu3JDtQq3rEGoNiIEDB3LRRRexYcMGAB5++GGqqqq47777mD17NpMnT+ajjz7i2muvZdSoUTzwwAN4PB4yMzN5/vnn6devH1VVVdxxxx2sWrUKEeHee++lvLyc9evX8+ijjwJ2XYr8/HweeeSRDn284RBJAbEPGBK0PdjZF0wB8Kkxxgt8KSJbsQJjH1ZoBLdd2vQGxpgngCfAltpob0dVg1CU3sfVV1/NXXfddURAvPzyy7z77rvceeed9OnTh+LiYqZPn84ll1yCSMth7AkJCbz++utHtcvPzw+5rkOoNSBKS1t2kXo8HtxyQKWlpaxYsQIR4amnnuJXv/oVv/71r/nFL/5/e3cfW1V9x3H8/QGuFHUBFHkYJStTGAIVwYka3GJAmA8MlmW1IHEaNW7LRAduBjEhTANxYdlT4pxGRVjQLqsPI2TzASnDuYiAA6qiiI+Up3YVGGoCFb/74/wKh3puKaXtOXC/r+Sm9zzd++23Pfd7z++c8/vdS/fu3amurj60Xi6XY968eSxYsIBcLsfChQt58MEHjzd9LdKeBWINMEjSQKIP/CnAtU3WeQaYCiyU1Iuoyek94F1gvqSeYb0JRCez20XDQfMjCOeORzPf9NvLyJEjqa2tZfv27dTV1dGzZ0/69u3LjBkzWLVqFZ06dWLbtm3s2rWLvn37NvtaZsbs2bO/tN2KFSsSx3VIGgPiaAUi3rFeTU0N5eXl7NixgwMHDhwa32H58uVUVFQcWq9nz+gjcOzYsSxbtoxzzz2XhoYGSktb3NByXNqtQJjZ55JuBZ4DOgOPmtkbku4B1prZ0rBsgqQ3gYPAL8ysHkDSvURFBuAeMzv6kEytizM6Se1HEM6dcMrKyqisrGTnzp2Ul5ezZMkS6urqWLduHblcjpKSkmbHUWjU2u3i4mM9AF/aPt5z6/Tp05k5cyaTJk1i5cqVzJ07t9nXvvnmm5k/fz5Dhgxp095aj6ZdPxXN7O9mNtjMzjazeWHenFAcsMhMMxtqZqVmVhHb9lEzOyc82q0jlIaDUcvUKZ39TmrnTjTl5eVUVFRQWVlJWVkZe/fupXfv3uRyOaqqqvjwww9b9Dr5tss3rkPSGBB9+vShtraW+vp69u/fz7Jly5p9v/79o+tuFi1adGj++PHjj+jqo/Go5KKLLmLr1q08/vjjTJ06taXpOW4F/7X5wMGo4vsRhHMnnmHDhrFv3z769+9Pv379mDZtGmvXrqW0tJTFixcfMW5Dc/Jtl29ch6QxIHK5HHPmzGH06NGMHz++2feeO3cuZWVlXHDBBYearyAa63r37t0MHz6cESNGUFVVdWjZNddcw5gxYw41O3WEgh8PYvenBxh57wvM/e5Qbhgz8OgbOOcAHw+io02cOJEZM2Ywbty4Vr/GsY4HUfBfmzt1Elef14+BZ/lNNM657NmzZw+DBw+mW7dux1UcWiPt+yBS171bjvuvHZV2GM65DlBdXc111113xLyuXbuyevXqlCI6uh49erB58+ZU3rvgC4RzrnCUlpayfv36tMM4YRR8E5NzrvVOlnOYhaA1fysvEM65VikqKqK+vt6LxAnAzKivr6eoqOiYtvMmJudcqxQXF1NTU0NdXV3aobgWKCoqori4+Ji28QLhnGuVXC53qIsId3LyJibnnHOJvEA455xL5AXCOedcopOmqw1JdUDLeuZK1gv4bxuF016yHmPW4wOPsa14jG0jCzF+zczOSlpw0hSI4yVpbb7+SLIi6zFmPT7wGNuKx9g2sh6jNzE555xL5AXCOedcIi8Qhz2UdgAtkPUYsx4feIxtxWNsG5mO0c9BOOecS+RHEM455xIVfIGQdIWktyVtkTQr7XgAJA2QVCXpTUlvSLo9zD9D0guS3gk/O27swfyxdpb0H0nLwvRASatDPv8i6ZSU4+shqVLSW5I2SbokS3mUNCP8jV+X9ISkoizkUNKjkmolvR6bl5g3Rf4Q4t0oqd0HWMkT34Lwd94o6WlJPWLL7grxvS3pO+0dX74YY8vukGSSeoXpDs9hSxR0gZDUGbgfuBIYCkyVNDTdqAD4HLjDzIYCFwM/DXHNAl40s0HAi2E6bbcDm2LTvwJ+a2bnALuBm1KJ6rDfA8+a2RBgBFGsmcijpP7AbcA3zWw40BmYQjZy+BhwRZN5+fJ2JTAoPG4BHkgpvheA4WZ2HrAZuAsg7DtTgGFhmz+GfT+NGJE0AJgAfBSbnUYOj6qgCwQwGthiZu+Z2QGgApicckyY2Q4zey0830f0odafKLZFYbVFwPfSiTAiqRi4Gng4TAsYC1SGVVKNUVJ34NvAIwBmdsDM9pCtPHYBuknqApwK7CADOTSzVcDHTWbny9tkYLFFXgF6SOrX0fGZ2fNm9nmYfAVo7Lp0MlBhZvvN7H1gC9G+367y5BDgt8CdQPwEcIfnsCUKvUD0B7bGpmvCvMyQVAKMBFYDfcxsR1i0E+iTUliNfkf0j/5FmD4T2BPbSdPO50CgDlgYmsEelnQaGcmjmW0Dfk30TXIHsBdYR7ZyGJcvb1ncj24E/hGeZyY+SZOBbWa2ocmizMQYV+gFItMknQ48CfzMzP4XX2bR5WepXYImaSJQa2br0oqhBboAo4AHzGwk8ClNmpPSzGNow59MVMi+CpxGQpNEFqX9/9ccSXcTNdMuSTuWOEmnArOBOWnH0lKFXiC2AQNi08VhXuok5YiKwxIzeyrM3tV42Bl+1qYVHzAGmCTpA6KmubFE7f09QnMJpJ/PGqDGzBpHpK8kKhhZyePlwPtmVmdmDcBTRHnNUg7j8uUtM/uRpBuAicA0O3wNf1biO5voy8CGsN8UA69J6kt2YjxCoReINcCgcNXIKUQnspamHFNjW/4jwCYz+01s0VLg+vD8euBvHR1bIzO7y8yKzayEKG8rzGwaUAX8IKyWdow7ga2SvhFmjQPeJDt5/Ai4WNKp4W/eGF9mcthEvrwtBX4YrsS5GNgba4rqMJKuIGrynGRmn8UWLQWmSOoqaSDRieBXOzo+M6s2s95mVhL2mxpgVPg/zUQOv8TMCvoBXEV0xcO7wN1pxxNiupTo8H0jsD48riJq438ReAdYDpyRdqwh3suAZeH514l2vi3AX4GuKcd2PrA25PIZoGeW8gj8EngLeB34M9A1CzkEniA6L9JA9EF2U768ASK6GvBdoJroqqw04ttC1I7fuM/8Kbb+3SG+t4Er08phk+UfAL3SymFLHn4ntXPOuUSF3sTknHMuDy8QzjnnEnmBcM45l8gLhHPOuUReIJxzziXyAuFcAkn/Dj9LJF3bxq89O+m9nMsav8zVuWZIugz4uZlNPIZtutjhvpSSln9iZqe3RXzOtSc/gnAugaRPwtP7gG9JWh/Gbugcxh1YE/rt/1FY/zJJL0laSnQ3NJKekbRO0XgPt4R59xH13rpe0pL4e4W7aBcoGhuiWlJ57LVX6vC4FkvCndfOtasuR1/FuYI2i9gRRPig32tmF0rqCrws6fmw7iii8QjeD9M3mtnHkroBayQ9aWazJN1qZucnvNf3ie78HgH0CtusCstGEo1nsB14majPpn+1/a/r3GF+BOHcsZlA1GfOeqIu2M8k6tsH4NVYcQC4TdIGorEJBsTWy+dS4AkzO2hmu4B/AhfGXrvGzL4g6kaipE1+G+ea4UcQzh0bAdPN7LkjZkbnKj5tMn05cImZfSZpJVB0HO+7P/b8IL7vug7gRxDONW8f8JXY9HPAT0J37EgaHAYhaqo7sDsUhyFEQ8c2amjcvomXgPJwnuMsotHwOrzXUeca+bcQ55q3ETgYmooeIxrzooSoH38RjViXNCTos8CPJW0i6kH0ldiyh4CNkl6zqIv0Rk8DlwAbiHrzvdPMdoYC41yH88tcnXPOJfImJuecc4m8QDjnnEvkBcI551wiLxDOOecSeYFwzjmXyAuEc865RF4gnHPOJfIC4ZxzLtH/AWI2EwTtPNclAAAAAElFTkSuQmCC\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":8,"id":"8623181e","metadata":{"id":"8623181e","executionInfo":{"status":"ok","timestamp":1669890668475,"user_tz":-420,"elapsed":13,"user":{"displayName":"Thanh Hùng","userId":"08335145449228425002"}}},"outputs":[],"source":[]},{"cell_type":"code","execution_count":8,"id":"c84dafda","metadata":{"id":"c84dafda","executionInfo":{"status":"ok","timestamp":1669890668475,"user_tz":-420,"elapsed":12,"user":{"displayName":"Thanh Hùng","userId":"08335145449228425002"}}},"outputs":[],"source":[]}],"metadata":{"colab":{"provenance":[],"toc_visible":true},"kernelspec":{"display_name":"Python 3 (ipykernel)","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":"3.8.0"},"toc":{"base_numbering":1,"nav_menu":{},"number_sections":true,"sideBar":true,"skip_h1_title":false,"title_cell":"Table of Contents","title_sidebar":"Contents","toc_cell":false,"toc_position":{},"toc_section_display":true,"toc_window_display":false},"gpuClass":"standard"},"nbformat":4,"nbformat_minor":5}