tensorflow2.0

czy36mengfei · Mar 11, 2019 · 6a32897ef1f72642047c31480abc0416c6f3ed9b · 6a32897
1 parent ebf68b0
commit 6a32897ef1f72642047c31480abc0416c6f3ed9b
Unified Split

Showing with 653 additions and 0 deletions.

+85 −0 004-keras_write_layers_from_scratch.ipynb

+123 −0 005-keras_saving_and_serializing_model.ipynb

+445 −0 101-example_text_classification.ipynb
diff --git a/004-keras_write_layers_from_scratch.ipynb b/004-keras_write_layers_from_scratch.ipynb
@@ -543,6 +543,91 @@
    "vae.fit(x_train, x_train, epochs=3, batch_size=64)"
   ]
  },
+  
+  {
+  
+   "cell_type": "markdown",
+  
+   "metadata": {},
+  
+   "source": [
+  
+    "自己编写训练方法"
+  
+   ]
+  
+  },
+  
+  {
+  
+   "cell_type": "code",
+  
+   "execution_count": 50,
+  
+   "metadata": {},
+  
+   "outputs": [
+  
+    {
+  
+     "name": "stdout",
+  
+     "output_type": "stream",
+  
+     "text": [
+  
+      "Start of epoch 0\n",
+  
+      "step 0: mean loss = tf.Tensor(213.26726, shape=(), dtype=float32)\n",
+  
+      "step 100: mean loss = tf.Tensor(6.5270114, shape=(), dtype=float32)\n",
+  
+      "step 200: mean loss = tf.Tensor(3.3300452, shape=(), dtype=float32)\n",
+  
+      "step 300: mean loss = tf.Tensor(2.2522914, shape=(), dtype=float32)\n",
+  
+      "step 400: mean loss = tf.Tensor(1.7097591, shape=(), dtype=float32)\n",
+  
+      "step 500: mean loss = tf.Tensor(1.3835965, shape=(), dtype=float32)\n",
+  
+      "step 600: mean loss = tf.Tensor(1.1659158, shape=(), dtype=float32)\n",
+  
+      "step 700: mean loss = tf.Tensor(1.0099952, shape=(), dtype=float32)\n",
+  
+      "step 800: mean loss = tf.Tensor(0.89288116, shape=(), dtype=float32)\n",
+  
+      "step 900: mean loss = tf.Tensor(0.8014734, shape=(), dtype=float32)\n",
+  
+      "Start of epoch 1\n",
+  
+      "step 0: mean loss = tf.Tensor(0.77191323, shape=(), dtype=float32)\n",
+  
+      "step 100: mean loss = tf.Tensor(0.70431703, shape=(), dtype=float32)\n",
+  
+      "step 200: mean loss = tf.Tensor(0.6486862, shape=(), dtype=float32)\n",
+  
+      "step 300: mean loss = tf.Tensor(0.60191154, shape=(), dtype=float32)\n",
+  
+      "step 400: mean loss = tf.Tensor(0.56213117, shape=(), dtype=float32)\n",
+  
+      "step 500: mean loss = tf.Tensor(0.52777255, shape=(), dtype=float32)\n",
+  
+      "step 600: mean loss = tf.Tensor(0.49796674, shape=(), dtype=float32)\n",
+  
+      "step 700: mean loss = tf.Tensor(0.47174037, shape=(), dtype=float32)\n",
+  
+      "step 800: mean loss = tf.Tensor(0.4485459, shape=(), dtype=float32)\n",
+  
+      "step 900: mean loss = tf.Tensor(0.4277973, shape=(), dtype=float32)\n",
+  
+      "Start of epoch 2\n",
+  
+      "step 0: mean loss = tf.Tensor(0.42051753, shape=(), dtype=float32)\n",
+  
+      "step 100: mean loss = tf.Tensor(0.40269083, shape=(), dtype=float32)\n",
+  
+      "step 200: mean loss = tf.Tensor(0.38661462, shape=(), dtype=float32)\n",
+  
+      "step 300: mean loss = tf.Tensor(0.3719676, shape=(), dtype=float32)\n",
+  
+      "step 400: mean loss = tf.Tensor(0.35864368, shape=(), dtype=float32)\n",
+  
+      "step 500: mean loss = tf.Tensor(0.3463759, shape=(), dtype=float32)\n",
+  
+      "step 600: mean loss = tf.Tensor(0.33514142, shape=(), dtype=float32)\n",
+  
+      "step 700: mean loss = tf.Tensor(0.3247494, shape=(), dtype=float32)\n",
+  
+      "step 800: mean loss = tf.Tensor(0.3151487, shape=(), dtype=float32)\n",
+  
+      "step 900: mean loss = tf.Tensor(0.3061987, shape=(), dtype=float32)\n"
+  
+     ]
+  
+    }
+  
+   ],
+  
+   "source": [
+  
+    "train_dataset = tf.data.Dataset.from_tensor_slices(x_train)\n",
+  
+    "train_dataset = train_dataset.shuffle(buffer_size=1024).batch(64)\n",
+  
+    "\n",
+  
+    "original_dim = 784\n",
+  
+    "vae = VAE(original_dim, 64, 32)\n",
+  
+    "\n",
+  
+    "optimizer = tf.keras.optimizers.Adam(learning_rate=1e-3)\n",
+  
+    "mse_loss_fn = tf.keras.losses.MeanSquaredError()\n",
+  
+    "\n",
+  
+    "loss_metric = tf.keras.metrics.Mean()\n",
+  
+    "\n",
+  
+    "# Iterate over epochs.\n",
+  
+    "for epoch in range(3):\n",
+  
+    "  print('Start of epoch %d' % (epoch,))\n",
+  
+    "\n",
+  
+    "  # Iterate over the batches of the dataset.\n",
+  
+    "  for step, x_batch_train in enumerate(train_dataset):\n",
+  
+    "    with tf.GradientTape() as tape:\n",
+  
+    "      reconstructed = vae(x_batch_train)\n",
+  
+    "      # Compute reconstruction loss\n",
+  
+    "      loss = mse_loss_fn(x_batch_train, reconstructed)\n",
+  
+    "      loss += sum(vae.losses)  # Add KLD regularization loss\n",
+  
+    "      \n",
+  
+    "    grads = tape.gradient(loss, vae.trainable_variables)\n",
+  
+    "    optimizer.apply_gradients(zip(grads, vae.trainable_variables))\n",
+  
+    "    \n",
+  
+    "    loss_metric(loss)\n",
+  
+    "    \n",
+  
+    "    if step % 100 == 0:\n",
+  
+    "      print('step %s: mean loss = %s' % (step, loss_metric.result()))"
+  
+   ]
+  
+  },
  {
   "cell_type": "code",
   "execution_count": null,

diff --git a/005-keras_saving_and_serializing_model.ipynb b/005-keras_saving_and_serializing_model.ipynb
@@ -0,0 +1,123 @@
+  
+{
+  
+ "cells": [
+  
+  {
+  
+   "cell_type": "markdown",
+  
+   "metadata": {},
+  
+   "source": [
+  
+    "# tensorflow2教程-keras模型保持和序列化\n",
+  
+    "\n",
+  
+    "## 1.保持序列模型和函数模型"
+  
+   ]
+  
+  },
+  
+  {
+  
+   "cell_type": "code",
+  
+   "execution_count": 4,
+  
+   "metadata": {},
+  
+   "outputs": [
+  
+    {
+  
+     "name": "stdout",
+  
+     "output_type": "stream",
+  
+     "text": [
+  
+      "Model: \"3_layer_mlp\"\n",
+  
+      "_________________________________________________________________\n",
+  
+      "Layer (type)                 Output Shape              Param #   \n",
+  
+      "=================================================================\n",
+  
+      "digits (InputLayer)          [(None, 784)]             0         \n",
+  
+      "_________________________________________________________________\n",
+  
+      "dense_1 (Dense)              (None, 64)                50240     \n",
+  
+      "_________________________________________________________________\n",
+  
+      "dense_2 (Dense)              (None, 64)                4160      \n",
+  
+      "_________________________________________________________________\n",
+  
+      "predictions (Dense)          (None, 10)                650       \n",
+  
+      "=================================================================\n",
+  
+      "Total params: 55,050\n",
+  
+      "Trainable params: 55,050\n",
+  
+      "Non-trainable params: 0\n",
+  
+      "_________________________________________________________________\n",
+  
+      "60000/60000 [==============================] - 1s 23us/sample - loss: 0.3136\n"
+  
+     ]
+  
+    }
+  
+   ],
+  
+   "source": [
+  
+    "# 构建一个简单的模型并训练\n",
+  
+    "from __future__ import absolute_import, division, print_function\n",
+  
+    "import tensorflow as tf\n",
+  
+    "tf.keras.backend.clear_session()\n",
+  
+    "from tensorflow import keras\n",
+  
+    "from tensorflow.keras import layers\n",
+  
+    "\n",
+  
+    "inputs = keras.Input(shape=(784,), name='digits')\n",
+  
+    "x = layers.Dense(64, activation='relu', name='dense_1')(inputs)\n",
+  
+    "x = layers.Dense(64, activation='relu', name='dense_2')(x)\n",
+  
+    "outputs = layers.Dense(10, activation='softmax', name='predictions')(x)\n",
+  
+    "\n",
+  
+    "model = keras.Model(inputs=inputs, outputs=outputs, name='3_layer_mlp')\n",
+  
+    "model.summary()\n",
+  
+    "(x_train, y_train), (x_test, y_test) = keras.datasets.mnist.load_data()\n",
+  
+    "x_train = x_train.reshape(60000, 784).astype('float32') / 255\n",
+  
+    "x_test = x_test.reshape(10000, 784).astype('float32') / 255\n",
+  
+    "\n",
+  
+    "model.compile(loss='sparse_categorical_crossentropy',\n",
+  
+    "              optimizer=keras.optimizers.RMSprop())\n",
+  
+    "history = model.fit(x_train, y_train,\n",
+  
+    "                    batch_size=64,\n",
+  
+    "                    epochs=1)\n",
+  
+    "\n",
+  
+    "predictions = model.predict(x_test)"
+  
+   ]
+  
+  },
+  
+  {
+  
+   "cell_type": "markdown",
+  
+   "metadata": {},
+  
+   "source": [
+  
+    "### 1.1保持全模型\n",
+  
+    "可以对整个模型进行保存，其保持的内容包括：\n",
+  
+    "- 该模型的架构\n",
+  
+    "- 模型的权重（在训练期间学到的）\n",
+  
+    "- 模型的训练配置（你传递给编译的），如果有的话\n",
+  
+    "- 优化器及其状态（如果有的话）（这使您可以从中断的地方重新启动训练）\n"
+  
+   ]
+  
+  },
+  
+  {
+  
+   "cell_type": "code",
+  
+   "execution_count": 5,
+  
+   "metadata": {},
+  
+   "outputs": [],
+  
+   "source": [
+  
+    "import numpy as np\n",
+  
+    "model.save('the_save_model.h5')\n",
+  
+    "new_model = keras.models.load_model('the_save_model.h5')\n",
+  
+    "new_prediction = new_model.predict(x_test)\n",
+  
+    "np.testing.assert_allclose(predictions, new_prediction, atol=1e-6) # 预测结果一样"
+  
+   ]
+  
+  },
+  
+  {
+  
+   "cell_type": "code",
+  
+   "execution_count": null,
+  
+   "metadata": {},
+  
+   "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": "3.6.8"
+  
+  }
+  
+ },
+  
+ "nbformat": 4,
+  
+ "nbformat_minor": 2
+  
+}