Skip to content

Commit

Permalink
Merge pull request #2324 from sanskarmodi8:issue#75196-fix
Browse files Browse the repository at this point in the history 
PiperOrigin-RevId: 673591318
  • Loading branch information
@copybara-github
copybara-github committed Sep 12, 2024
2 parents 45a981f + b90b419 commit 7df1042
Showing 1 changed file with 28 additions and 58 deletions.
86 changes: 28 additions & 58 deletions site/en/tutorials/keras/save_and_load.ipynb
View file
Original file line number Diff line number Diff line change
Expand Up @@ -142,7 +142,7 @@
},
"outputs": [],
"source": [
"!pip install pyyaml h5py # Required to save models in HDF5 format"
"!pip install pyyaml h5py # Required to save models in HDF5 format."
]
},
{
Expand Down Expand Up @@ -266,7 +266,7 @@
},
"outputs": [],
"source": [
"checkpoint_path = \"training_1/cp.ckpt\"\n",
"checkpoint_path = \"training_1/cp.weights.h5\" # we are only saving weights therefore we need to use .weights.h5 extension instead we would use .keras for whole model\n",
"checkpoint_dir = os.path.dirname(checkpoint_path)\n",
"\n",
"# Create a callback that saves the model's weights\n",
Expand All @@ -275,8 +275,8 @@
" verbose=1)\n",
"\n",
"# Train the model with the new callback\n",
"model.fit(train_images, \n",
" train_labels, \n",
"model.fit(train_images,\n",
" train_labels,\n",
" epochs=10,\n",
" validation_data=(test_images, test_labels),\n",
" callbacks=[cp_callback]) # Pass callback to training\n",
Expand Down Expand Up @@ -312,7 +312,7 @@
"id": "wlRN_f56Pqa9"
},
"source": [
"As long as two models share the same architecture you can share weights between them. So, when restoring a model from weights-only, create a model with the same architecture as the original model and then set its weights. \n",
"As long as two models share the same architecture you can share weights between them. So, when restoring a model from weights-only, create a model with the same architecture as the original model and then set its weights.\n",
"\n",
"Now rebuild a fresh, untrained model and evaluate it on the test set. An untrained model will perform at chance levels (~10% accuracy):"
]
Expand Down Expand Up @@ -380,8 +380,9 @@
"outputs": [],
"source": [
"# Include the epoch in the file name (uses `str.format`)\n",
"checkpoint_path = \"training_2/cp-{epoch:04d}.ckpt\"\n",
"checkpoint_path = \"training_2/cp-{epoch:04d}.weights.h5\"\n",
"checkpoint_dir = os.path.dirname(checkpoint_path)\n",
"os.mkdir(checkpoint_dir)\n",
"\n",
"batch_size = 32\n",
"\n",
Expand All @@ -392,8 +393,8 @@
"\n",
"# Create a callback that saves the model's weights every 5 epochs\n",
"cp_callback = tf.keras.callbacks.ModelCheckpoint(\n",
" filepath=checkpoint_path, \n",
" verbose=1, \n",
" filepath=checkpoint_path,\n",
" verbose=1,\n",
" save_weights_only=True,\n",
" save_freq=5*n_batches)\n",
"\n",
Expand All @@ -404,10 +405,10 @@
"model.save_weights(checkpoint_path.format(epoch=0))\n",
"\n",
"# Train the model with the new callback\n",
"model.fit(train_images, \n",
"model.fit(train_images,\n",
" train_labels,\n",
" epochs=50, \n",
" batch_size=batch_size, \n",
" epochs=50,\n",
" batch_size=batch_size,\n",
" callbacks=[cp_callback],\n",
" validation_data=(test_images, test_labels),\n",
" verbose=0)"
Expand Down Expand Up @@ -441,7 +442,11 @@
},
"outputs": [],
"source": [
"latest = tf.train.latest_checkpoint(checkpoint_dir)\n",
"def load_latest_checkpoint(checkpoint_dir):\n",
" latest = max(os.listdir(checkpoint_dir), key=lambda f: int(f.split('-')[1].split('.')[0]))\n",
" return os.path.join(checkpoint_dir, latest)\n",
"\n",
"latest = load_latest_checkpoint(checkpoint_dir)\n",
"latest"
]
},
Expand Down Expand Up @@ -505,7 +510,7 @@
"source": [
"## Manually save weights\n",
"\n",
"To save weights manually, use `tf.keras.Model.save_weights`. By default, `tf.keras`—and the `Model.save_weights` method in particular—uses the TensorFlow [Checkpoint](../../guide/checkpoint.ipynb) format with a `.ckpt` extension. To save in the HDF5 format with a `.h5` extension, refer to the [Save and load models](https://www.tensorflow.org/guide/keras/save_and_serialize) guide."
"To save weights manually, use `tf.keras.Model.save_weights`. You have to use .weights.h5 extension to save the weights. You can refer to the [Save and load models](https://www.tensorflow.org/guide/keras/save_and_serialize) guide."
]
},
{
Expand All @@ -517,13 +522,14 @@
"outputs": [],
"source": [
"# Save the weights\n",
"model.save_weights('./checkpoints/my_checkpoint')\n",
"os.mkdir('./checkpoints')\n",
"model.save_weights('./checkpoints/my_checkpoint.weights.h5')\n",
"\n",
"# Create a new model instance\n",
"model = create_model()\n",
"\n",
"# Restore the weights\n",
"model.load_weights('./checkpoints/my_checkpoint')\n",
"model.load_weights('./checkpoints/my_checkpoint.weights.h5')\n",
"\n",
"# Evaluate the model\n",
"loss, acc = model.evaluate(test_images, test_labels, verbose=2)\n",
Expand Down Expand Up @@ -657,7 +663,7 @@
"id": "LtcN4VIb7JkK"
},
"source": [
"The SavedModel format is another way to serialize models. Models saved in this format can be restored using `tf.keras.models.load_model` and are compatible with TensorFlow Serving. The [SavedModel guide](../../guide/saved_model.ipynb) goes into detail about how to `serve/inspect` the SavedModel. The section below illustrates the steps to save and restore the model."
"The SavedModel format is another way to serialize models. Models saved in this format can directly be used with TFLite/TFServing/etc for inferencing. The [SavedModel guide](../../guide/saved_model.ipynb) goes into detail about how to `serve/inspect` the SavedModel. The section below illustrates the steps to save and restore the model."
]
},
{
Expand All @@ -674,7 +680,7 @@
"\n",
"# Save the entire model as a SavedModel.\n",
"!mkdir -p saved_model\n",
"model.save('saved_model/my_model') "
"tf.saved_model.save(model, 'saved_model/my_model')"
]
},
{
Expand Down Expand Up @@ -707,7 +713,7 @@
"id": "B7qfpvpY9HCe"
},
"source": [
"Reload a fresh Keras model from the saved model:"
"Reload the saved SavedModel file:"
]
},
{
Expand All @@ -718,34 +724,8 @@
},
"outputs": [],
"source": [
"new_model = tf.keras.models.load_model('saved_model/my_model')\n",
"\n",
"# Check its architecture\n",
"new_model.summary()"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "uWwgNaz19TH2"
},
"source": [
"The restored model is compiled with the same arguments as the original model. Try running evaluate and predict with the loaded model:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"id": "Yh5Mu0yOgE5J"
},
"outputs": [],
"source": [
"# Evaluate the restored model\n",
"loss, acc = new_model.evaluate(test_images, test_labels, verbose=2)\n",
"print('Restored model, accuracy: {:5.2f}%'.format(100 * acc))\n",
"\n",
"print(new_model.predict(test_images).shape)"
"saved_model = tf.saved_model.load('saved_model/my_model')\n",
"saved_model"
]
},
{
Expand All @@ -756,7 +736,7 @@
"source": [
"### HDF5 format\n",
"\n",
"Keras provides a basic legacy high-level save format using the [HDF5](https://en.wikipedia.org/wiki/Hierarchical_Data_Format) standard. "
"Keras provides a basic legacy high-level save format using the [HDF5](https://en.wikipedia.org/wiki/Hierarchical_Data_Format) standard."
]
},
{
Expand All @@ -773,7 +753,7 @@
"\n",
"# Save the entire model to a HDF5 file.\n",
"# The '.h5' extension indicates that the model should be saved to HDF5.\n",
"model.save('my_model.h5') "
"model.save('my_model.h5')"
]
},
{
Expand Down Expand Up @@ -859,21 +839,11 @@
"\n",
"Refer to the [Writing layers and models from scratch](https://www.tensorflow.org/guide/keras/custom_layers_and_models) tutorial for examples of custom objects and `get_config`.\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"id": "jBVTkkUIkEF3"
},
"outputs": [],
"source": []
}
],
"metadata": {
"accelerator": "GPU",
"colab": {
"collapsed_sections": [],
"name": "save_and_load.ipynb",
"toc_visible": true
},
Expand Down

0 comments on commit 7df1042

Please sign in to comment.