0

我正在尝试在 ResNet50 上执行模型修剪。但是，我对它们的测试精度完全相同。我究竟做错了什么？

import tensorflow_model_optimization as tfmot
import numpy as np

prune_low_magnitude = tfmot.sparsity.keras.prune_low_magnitude

# Compute end step to finish pruning after 2 epochs.
batch_size = 128
epochs = 50
validation_split = 0.1 # 10% of training set will be used for validation set. 

num_images = x_train.shape[0] * (1 - validation_split)
end_step = np.ceil(num_images / batch_size).astype(np.int32) * epochs

# Define model for pruning.
pruning_params = {
      'pruning_schedule': tfmot.sparsity.keras.PolynomialDecay(initial_sparsity=0.50,
                                                               final_sparsity=0.80,
                                                               begin_step=0,
                                                               end_step=end_step)
}

model_for_pruning = prune_low_magnitude(model, **pruning_params)

# `prune_low_magnitude` requires a recompile.
model_for_pruning.compile(optimizer='adam',
              loss=tf.keras.losses.CategoricalCrossentropy(from_logits=True),
              metrics=['accuracy'])

import tempfile
import os
logdir = tempfile.mkdtemp()

callbacks = [
  tfmot.sparsity.keras.UpdatePruningStep(),
  tfmot.sparsity.keras.PruningSummaries(log_dir=logdir),
]

model_for_pruning.fit(x_train, y_train,
                  batch_size=batch_size, epochs=5, validation_split=validation_split,
                  verbose = 1,
                  callbacks=callbacks)

_, baseline_model_accuracy = model.evaluate(x_test, y_test, verbose=0)
_, model_for_pruning_accuracy = model_for_pruning.evaluate(x_test, y_test, verbose=0)

print('Baseline test accuracy:', baseline_model_accuracy) 
print('Pruned test accuracy:', model_for_pruning_accuracy)

输出：基线测试精度：0.9070000052452087 修剪后的测试精度：0.9070000052452087

4

0 回答 0