![]() |
![]() |
![]() |
![]() |
Overview
This is an end to end example showing the usage of the cluster preserving quantization aware training (CQAT) API, part of the TensorFlow Model Optimization Toolkit's collaborative optimization pipeline.
Other pages
For an introduction to the pipeline and other available techniques, see the collaborative optimization overview page.
Contents
In the tutorial, you will:
- Train a
tf.keras
model for the MNIST dataset from scratch. - Fine-tune the model with clustering and see the accuracy.
- Apply QAT and observe the loss of clusters.
- Apply CQAT and observe that the clustering applied earlier has been preserved.
- Generate a TFLite model and observe the effects of applying CQAT on it.
- Compare the achieved CQAT model accuracy with a model quantized using post-training quantization.
Setup
You can run this Jupyter Notebook in your local virtualenv or colab. For details of setting up dependencies, please refer to the installation guide.
pip install -q tensorflow-model-optimization
import tensorflow as tf
import numpy as np
import tempfile
import zipfile
import os
Train a tf.keras model for MNIST without clustering
# Load MNIST dataset
mnist = tf.keras.datasets.mnist
(train_images, train_labels), (test_images, test_labels) = mnist.load_data()
# Normalize the input image so that each pixel value is between 0 to 1.
train_images = train_images / 255.0
test_images = test_images / 255.0
model = tf.keras.Sequential([
tf.keras.layers.InputLayer(input_shape=(28, 28)),
tf.keras.layers.Reshape(target_shape=(28, 28, 1)),
tf.keras.layers.Conv2D(filters=12, kernel_size=(3, 3),
activation=tf.nn.relu),
tf.keras.layers.MaxPooling2D(pool_size=(2, 2)),
tf.keras.layers.Flatten(),
tf.keras.layers.Dense(10)
])
# Train the digit classification model
model.compile(optimizer='adam',
loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),
metrics=['accuracy'])
model.fit(
train_images,
train_labels,
validation_split=0.1,
epochs=10
)
2023-05-26 11:40:29.139361: E tensorflow/compiler/xla/stream_executor/cuda/cuda_driver.cc:268] failed call to cuInit: CUDA_ERROR_NO_DEVICE: no CUDA-capable device is detected Epoch 1/10 1688/1688 [==============================] - 8s 4ms/step - loss: 0.2971 - accuracy: 0.9176 - val_loss: 0.1145 - val_accuracy: 0.9703 Epoch 2/10 1688/1688 [==============================] - 7s 4ms/step - loss: 0.1141 - accuracy: 0.9676 - val_loss: 0.0821 - val_accuracy: 0.9767 Epoch 3/10 1688/1688 [==============================] - 7s 4ms/step - loss: 0.0823 - accuracy: 0.9760 - val_loss: 0.0679 - val_accuracy: 0.9823 Epoch 4/10 1688/1688 [==============================] - 7s 4ms/step - loss: 0.0685 - accuracy: 0.9801 - val_loss: 0.0648 - val_accuracy: 0.9835 Epoch 5/10 1688/1688 [==============================] - 7s 4ms/step - loss: 0.0593 - accuracy: 0.9825 - val_loss: 0.0677 - val_accuracy: 0.9818 Epoch 6/10 1688/1688 [==============================] - 7s 4ms/step - loss: 0.0528 - accuracy: 0.9841 - val_loss: 0.0571 - val_accuracy: 0.9847 Epoch 7/10 1688/1688 [==============================] - 7s 4ms/step - loss: 0.0473 - accuracy: 0.9856 - val_loss: 0.0611 - val_accuracy: 0.9830 Epoch 8/10 1688/1688 [==============================] - 7s 4ms/step - loss: 0.0431 - accuracy: 0.9871 - val_loss: 0.0645 - val_accuracy: 0.9835 Epoch 9/10 1688/1688 [==============================] - 7s 4ms/step - loss: 0.0394 - accuracy: 0.9878 - val_loss: 0.0590 - val_accuracy: 0.9858 Epoch 10/10 1688/1688 [==============================] - 7s 4ms/step - loss: 0.0359 - accuracy: 0.9894 - val_loss: 0.0588 - val_accuracy: 0.9845 <keras.src.callbacks.History at 0x7fd5db64ca90>
Evaluate the baseline model and save it for later usage
_, baseline_model_accuracy = model.evaluate(
test_images, test_labels, verbose=0)
print('Baseline test accuracy:', baseline_model_accuracy)
_, keras_file = tempfile.mkstemp('.h5')
print('Saving model to: ', keras_file)
tf.keras.models.save_model(model, keras_file, include_optimizer=False)
Baseline test accuracy: 0.9810000061988831 Saving model to: /tmpfs/tmp/tmp7vdv4xcl.h5 /tmpfs/tmp/ipykernel_40067/1025833175.py:8: UserWarning: You are saving your model as an HDF5 file via `model.save()`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')`. tf.keras.models.save_model(model, keras_file, include_optimizer=False)
Cluster and fine-tune the model with 8 clusters
Apply the cluster_weights()
API to cluster the whole pre-trained model to demonstrate and observe its effectiveness in reducing the model size when applying zip, while maintaining accuracy. For how best to use the API to achieve the best compression rate while maintaining your target accuracy, refer to the clustering comprehensive guide.
Define the model and apply the clustering API
The model needs to be pre-trained before using the clustering API.
import tensorflow_model_optimization as tfmot
cluster_weights = tfmot.clustering.keras.cluster_weights
CentroidInitialization = tfmot.clustering.keras.CentroidInitialization
clustering_params = {
'number_of_clusters': 8,
'cluster_centroids_init': CentroidInitialization.KMEANS_PLUS_PLUS,
'cluster_per_channel': True,
}
clustered_model = cluster_weights(model, **clustering_params)
# Use smaller learning rate for fine-tuning
opt = tf.keras.optimizers.Adam(learning_rate=1e-5)
clustered_model.compile(
loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),
optimizer=opt,
metrics=['accuracy'])
clustered_model.summary()
Model: "sequential" _________________________________________________________________ Layer (type) Output Shape Param # ================================================================= cluster_reshape (ClusterWe (None, 28, 28, 1) 0 ights) cluster_conv2d (ClusterWei (None, 26, 26, 12) 324 ghts) cluster_max_pooling2d (Clu (None, 13, 13, 12) 0 sterWeights) cluster_flatten (ClusterWe (None, 2028) 0 ights) cluster_dense (ClusterWeig (None, 10) 40578 hts) ================================================================= Total params: 40902 (239.41 KB) Trainable params: 20514 (80.13 KB) Non-trainable params: 20388 (159.28 KB) _________________________________________________________________
Fine-tune the model and evaluate the accuracy against baseline
Fine-tune the model with clustering for 3 epochs.
# Fine-tune model
clustered_model.fit(
train_images,
train_labels,
epochs=3,
validation_split=0.1)
Epoch 1/3 1688/1688 [==============================] - 10s 5ms/step - loss: 0.0346 - accuracy: 0.9895 - val_loss: 0.0549 - val_accuracy: 0.9847 Epoch 2/3 1688/1688 [==============================] - 8s 5ms/step - loss: 0.0311 - accuracy: 0.9913 - val_loss: 0.0529 - val_accuracy: 0.9847 Epoch 3/3 1688/1688 [==============================] - 8s 5ms/step - loss: 0.0299 - accuracy: 0.9918 - val_loss: 0.0522 - val_accuracy: 0.9847 <keras.src.callbacks.History at 0x7fd5dbb5b760>
Define helper functions to calculate and print the number of clustering in each kernel of the model.
def print_model_weight_clusters(model):
for layer in model.layers:
if isinstance(layer, tf.keras.layers.Wrapper):
weights = layer.trainable_weights
else:
weights = layer.weights
for weight in weights:
# ignore auxiliary quantization weights
if "quantize_layer" in weight.name:
continue
if "kernel" in weight.name:
unique_count = len(np.unique(weight))
print(
f"{layer.name}/{weight.name}: {unique_count} clusters "
)
Check that the model kernels were correctly clustered. We need to strip the clustering wrapper first.
stripped_clustered_model = tfmot.clustering.keras.strip_clustering(clustered_model)
print_model_weight_clusters(stripped_clustered_model)
conv2d/kernel:0: 96 clusters dense/kernel:0: 8 clusters
For this example, there is minimal loss in test accuracy after clustering, compared to the baseline.
_, clustered_model_accuracy = clustered_model.evaluate(
test_images, test_labels, verbose=0)
print('Baseline test accuracy:', baseline_model_accuracy)
print('Clustered test accuracy:', clustered_model_accuracy)
Baseline test accuracy: 0.9810000061988831 Clustered test accuracy: 0.982200026512146
Apply QAT and CQAT and check effect on model clusters in both cases
Next, we apply both QAT and cluster preserving QAT (CQAT) on the clustered model and observe that CQAT preserves weight clusters in your clustered model. Note that we stripped clustering wrappers from your model with tfmot.clustering.keras.strip_clustering
before applying CQAT API.
# QAT
qat_model = tfmot.quantization.keras.quantize_model(stripped_clustered_model)
qat_model.compile(optimizer='adam',
loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),
metrics=['accuracy'])
print('Train qat model:')
qat_model.fit(train_images, train_labels, batch_size=128, epochs=1, validation_split=0.1)
# CQAT
quant_aware_annotate_model = tfmot.quantization.keras.quantize_annotate_model(
stripped_clustered_model)
cqat_model = tfmot.quantization.keras.quantize_apply(
quant_aware_annotate_model,
tfmot.experimental.combine.Default8BitClusterPreserveQuantizeScheme())
cqat_model.compile(optimizer='adam',
loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),
metrics=['accuracy'])
print('Train cqat model:')
cqat_model.fit(train_images, train_labels, batch_size=128, epochs=1, validation_split=0.1)
Train qat model: 422/422 [==============================] - 4s 7ms/step - loss: 0.0320 - accuracy: 0.9910 - val_loss: 0.0532 - val_accuracy: 0.9852 WARNING:root:Input layer does not contain zero weights, so apply CQAT instead. WARNING:root:Input layer does not contain zero weights, so apply CQAT instead. Train cqat model: WARNING:tensorflow:Gradients do not exist for variables ['conv2d/kernel:0', 'dense/kernel:0'] when minimizing the loss. If you're using `model.compile()`, did you forget to provide a `loss` argument? WARNING:tensorflow:Gradients do not exist for variables ['conv2d/kernel:0', 'dense/kernel:0'] when minimizing the loss. If you're using `model.compile()`, did you forget to provide a `loss` argument? WARNING:tensorflow:Gradients do not exist for variables ['conv2d/kernel:0', 'dense/kernel:0'] when minimizing the loss. If you're using `model.compile()`, did you forget to provide a `loss` argument? WARNING:tensorflow:Gradients do not exist for variables ['conv2d/kernel:0', 'dense/kernel:0'] when minimizing the loss. If you're using `model.compile()`, did you forget to provide a `loss` argument? WARNING:tensorflow:Gradients do not exist for variables ['conv2d/kernel:0', 'dense/kernel:0'] when minimizing the loss. If you're using `model.compile()`, did you forget to provide a `loss` argument? WARNING:tensorflow:Gradients do not exist for variables ['conv2d/kernel:0', 'dense/kernel:0'] when minimizing the loss. If you're using `model.compile()`, did you forget to provide a `loss` argument? WARNING:tensorflow:Gradients do not exist for variables ['conv2d/kernel:0', 'dense/kernel:0'] when minimizing the loss. If you're using `model.compile()`, did you forget to provide a `loss` argument? WARNING:tensorflow:Gradients do not exist for variables ['conv2d/kernel:0', 'dense/kernel:0'] when minimizing the loss. If you're using `model.compile()`, did you forget to provide a `loss` argument? 422/422 [==============================] - 5s 7ms/step - loss: 0.0290 - accuracy: 0.9916 - val_loss: 0.0557 - val_accuracy: 0.9852 <keras.src.callbacks.History at 0x7fd5db4560a0>
print("QAT Model clusters:")
print_model_weight_clusters(qat_model)
print("CQAT Model clusters:")
print_model_weight_clusters(cqat_model)
QAT Model clusters: quant_conv2d/conv2d/kernel:0: 108 clusters quant_dense/dense/kernel:0: 19955 clusters CQAT Model clusters: quant_conv2d/conv2d/kernel:0: 96 clusters quant_dense/dense/kernel:0: 8 clusters
See compression benefits of CQAT model
Define helper function to get zipped model file.
def get_gzipped_model_size(file):
# It returns the size of the gzipped model in kilobytes.
_, zipped_file = tempfile.mkstemp('.zip')
with zipfile.ZipFile(zipped_file, 'w', compression=zipfile.ZIP_DEFLATED) as f:
f.write(file)
return os.path.getsize(zipped_file)/1000
Note that this is a small model. Applying clustering and CQAT to a bigger production model would yield a more significant compression.
# QAT model
converter = tf.lite.TFLiteConverter.from_keras_model(qat_model)
converter.optimizations = [tf.lite.Optimize.DEFAULT]
qat_tflite_model = converter.convert()
qat_model_file = 'qat_model.tflite'
# Save the model.
with open(qat_model_file, 'wb') as f:
f.write(qat_tflite_model)
# CQAT model
converter = tf.lite.TFLiteConverter.from_keras_model(cqat_model)
converter.optimizations = [tf.lite.Optimize.DEFAULT]
cqat_tflite_model = converter.convert()
cqat_model_file = 'cqat_model.tflite'
# Save the model.
with open(cqat_model_file, 'wb') as f:
f.write(cqat_tflite_model)
print("QAT model size: ", get_gzipped_model_size(qat_model_file), ' KB')
print("CQAT model size: ", get_gzipped_model_size(cqat_model_file), ' KB')
INFO:tensorflow:Assets written to: /tmpfs/tmp/tmpd6i99pdt/assets INFO:tensorflow:Assets written to: /tmpfs/tmp/tmpd6i99pdt/assets /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/lite/python/convert.py:887: UserWarning: Statistics for quantized inputs were expected, but not specified; continuing anyway. warnings.warn( 2023-05-26 11:42:18.919068: W tensorflow/compiler/mlir/lite/python/tf_tfl_flatbuffer_helpers.cc:364] Ignored output_format. 2023-05-26 11:42:18.919107: W tensorflow/compiler/mlir/lite/python/tf_tfl_flatbuffer_helpers.cc:367] Ignored drop_control_dependency. INFO:tensorflow:Assets written to: /tmpfs/tmp/tmp58obfmbo/assets INFO:tensorflow:Assets written to: /tmpfs/tmp/tmp58obfmbo/assets QAT model size: 17.777 KB CQAT model size: 11.15 KB /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/lite/python/convert.py:887: UserWarning: Statistics for quantized inputs were expected, but not specified; continuing anyway. warnings.warn( 2023-05-26 11:42:21.724626: W tensorflow/compiler/mlir/lite/python/tf_tfl_flatbuffer_helpers.cc:364] Ignored output_format. 2023-05-26 11:42:21.724672: W tensorflow/compiler/mlir/lite/python/tf_tfl_flatbuffer_helpers.cc:367] Ignored drop_control_dependency.
See the persistence of accuracy from TF to TFLite
Define a helper function to evaluate the TFLite model on the test dataset.
def eval_model(interpreter):
input_index = interpreter.get_input_details()[0]["index"]
output_index = interpreter.get_output_details()[0]["index"]
# Run predictions on every image in the "test" dataset.
prediction_digits = []
for i, test_image in enumerate(test_images):
if i % 1000 == 0:
print(f"Evaluated on {i} results so far.")
# Pre-processing: add batch dimension and convert to float32 to match with
# the model's input data format.
test_image = np.expand_dims(test_image, axis=0).astype(np.float32)
interpreter.set_tensor(input_index, test_image)
# Run inference.
interpreter.invoke()
# Post-processing: remove batch dimension and find the digit with highest
# probability.
output = interpreter.tensor(output_index)
digit = np.argmax(output()[0])
prediction_digits.append(digit)
print('\n')
# Compare prediction results with ground truth labels to calculate accuracy.
prediction_digits = np.array(prediction_digits)
accuracy = (prediction_digits == test_labels).mean()
return accuracy
You evaluate the model, which has been clustered and quantized, and then see the accuracy from TensorFlow persists in the TFLite backend.
interpreter = tf.lite.Interpreter(cqat_model_file)
interpreter.allocate_tensors()
cqat_test_accuracy = eval_model(interpreter)
print('Clustered and quantized TFLite test_accuracy:', cqat_test_accuracy)
print('Clustered TF test accuracy:', clustered_model_accuracy)
INFO: Created TensorFlow Lite XNNPACK delegate for CPU. Evaluated on 0 results so far. Evaluated on 1000 results so far. Evaluated on 2000 results so far. Evaluated on 3000 results so far. Evaluated on 4000 results so far. Evaluated on 5000 results so far. Evaluated on 6000 results so far. Evaluated on 7000 results so far. Evaluated on 8000 results so far. Evaluated on 9000 results so far. Clustered and quantized TFLite test_accuracy: 0.9828 Clustered TF test accuracy: 0.982200026512146
Apply post-training quantization and compare to CQAT model
Next, we use post-training quantization (no fine-tuning) on the clustered model and check its accuracy against the CQAT model. This demonstrates why you would need to use CQAT to improve the quantized model's accuracy. The difference may not be very visible, because the MNIST model is quite small and overparametrized.
First, define a generator for the callibration dataset from the first 1000 training images.
def mnist_representative_data_gen():
for image in train_images[:1000]:
image = np.expand_dims(image, axis=0).astype(np.float32)
yield [image]
Quantize the model and compare accuracy to the previously acquired CQAT model. Note that the model quantized with fine-tuning achieves higher accuracy.
converter = tf.lite.TFLiteConverter.from_keras_model(stripped_clustered_model)
converter.optimizations = [tf.lite.Optimize.DEFAULT]
converter.representative_dataset = mnist_representative_data_gen
post_training_tflite_model = converter.convert()
post_training_model_file = 'post_training_model.tflite'
# Save the model.
with open(post_training_model_file, 'wb') as f:
f.write(post_training_tflite_model)
# Compare accuracy
interpreter = tf.lite.Interpreter(post_training_model_file)
interpreter.allocate_tensors()
post_training_test_accuracy = eval_model(interpreter)
print('CQAT TFLite test_accuracy:', cqat_test_accuracy)
print('Post-training (no fine-tuning) TF test accuracy:', post_training_test_accuracy)
INFO:tensorflow:Assets written to: /tmpfs/tmp/tmpgg_e6hrv/assets INFO:tensorflow:Assets written to: /tmpfs/tmp/tmpgg_e6hrv/assets /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/lite/python/convert.py:887: UserWarning: Statistics for quantized inputs were expected, but not specified; continuing anyway. warnings.warn( 2023-05-26 11:42:23.303658: W tensorflow/compiler/mlir/lite/python/tf_tfl_flatbuffer_helpers.cc:364] Ignored output_format. 2023-05-26 11:42:23.303696: W tensorflow/compiler/mlir/lite/python/tf_tfl_flatbuffer_helpers.cc:367] Ignored drop_control_dependency. fully_quantize: 0, inference_type: 6, input_inference_type: FLOAT32, output_inference_type: FLOAT32 Evaluated on 0 results so far. Evaluated on 1000 results so far. Evaluated on 2000 results so far. Evaluated on 3000 results so far. Evaluated on 4000 results so far. Evaluated on 5000 results so far. Evaluated on 6000 results so far. Evaluated on 7000 results so far. Evaluated on 8000 results so far. Evaluated on 9000 results so far. CQAT TFLite test_accuracy: 0.9828 Post-training (no fine-tuning) TF test accuracy: 0.982
Conclusion
In this tutorial, you learned how to create a model, cluster it using the cluster_weights()
API, and apply the cluster preserving quantization aware training (CQAT) to preserve clusters while using QAT. The final CQAT model was compared to the QAT one to show that the clusters are preserved in the former and lost in the latter. Next, the models were converted to TFLite to show the compression benefits of chaining clustering and CQAT model optimization techniques and the TFLite model was evaluated to ensure that the accuracy persists in the TFLite backend. Finally, the CQAT model was compared to a quantized clustered model achieved using the post-training quantization API to demonstrate the advantage of CQAT in recovering the accuracy loss from normal quantization.