Esta página foi traduzida pela API Cloud Translation.
Switch to English

Guia abrangente de poda

Ver no TensorFlow.org Executar no Google Colab Ver fonte no GitHub Baixar caderno

Bem-vindo ao guia completo para poda de peso Keras.

Esta página documenta vários casos de uso e mostra como usar a API para cada um. Depois de saber quais APIs você precisa, encontre os parâmetros e os detalhes de baixo nível nos documentos da API .

  • Se você quiser ver os benefícios da poda e o que é compatível, consulte a visão geral .
  • Para um único exemplo ponta a ponta, consulte o exemplo de poda .

Os seguintes casos de uso são cobertos:

  • Defina e treine um modelo podado.
    • Sequencial e funcional.
    • Keras model.fit e loops de treinamento personalizados
  • Faça checkpoint e desserialize um modelo podado.
  • Implante um modelo podado e veja os benefícios da compressão.

Para configuração do algoritmo de remoção, consulte a documentação da API tfmot.sparsity.keras.prune_low_magnitude .

Configuração

Para encontrar as APIs de que você precisa e entender os propósitos, você pode executar, mas pule a leitura desta seção.

! pip install -q tensorflow-model-optimization

import tensorflow as tf
import numpy as np
import tensorflow_model_optimization as tfmot

%load_ext tensorboard

import tempfile

input_shape = [20]
x_train = np.random.randn(1, 20).astype(np.float32)
y_train = tf.keras.utils.to_categorical(np.random.randn(1), num_classes=20)

def setup_model():
  model = tf.keras.Sequential([
      tf.keras.layers.Dense(20, input_shape=input_shape),
      tf.keras.layers.Flatten()
  ])
  return model

def setup_pretrained_weights():
  model = setup_model()

  model.compile(
      loss=tf.keras.losses.categorical_crossentropy,
      optimizer='adam',
      metrics=['accuracy']
  )

  model.fit(x_train, y_train)

  _, pretrained_weights = tempfile.mkstemp('.tf')

  model.save_weights(pretrained_weights)

  return pretrained_weights

def get_gzipped_model_size(model):
  # Returns size of gzipped model, in bytes.
  import os
  import zipfile

  _, keras_file = tempfile.mkstemp('.h5')
  model.save(keras_file, include_optimizer=False)

  _, zipped_file = tempfile.mkstemp('.zip')
  with zipfile.ZipFile(zipped_file, 'w', compression=zipfile.ZIP_DEFLATED) as f:
    f.write(keras_file)

  return os.path.getsize(zipped_file)

setup_model()
pretrained_weights = setup_pretrained_weights()
WARNING: You are using pip version 20.2.2; however, version 20.2.3 is available.
You should consider upgrading via the '/tmpfs/src/tf_docs_env/bin/python -m pip install --upgrade pip' command.
1/1 [==============================] - 0s 2ms/step - loss: 1.1999 - accuracy: 0.0000e+00

Definir modelo

Podar modelo inteiro (sequencial e funcional)

Dicas para melhor precisão do modelo:

  • Experimente "Prune some layers" para pular a poda das camadas que mais reduzem a precisão.
  • Geralmente, é melhor fazer o ajuste fino com a poda, em vez de treinar do zero.

Para fazer todo o modelo treinar com poda, aplique tfmot.sparsity.keras.prune_low_magnitude ao modelo.

base_model = setup_model()
base_model.load_weights(pretrained_weights) # optional but recommended.

model_for_pruning = tfmot.sparsity.keras.prune_low_magnitude(base_model)

model_for_pruning.summary()
WARNING:tensorflow:From /tmpfs/src/tf_docs_env/lib/python3.6/site-packages/tensorflow_model_optimization/python/core/sparsity/keras/pruning_wrapper.py:200: Layer.add_variable (from tensorflow.python.keras.engine.base_layer) is deprecated and will be removed in a future version.
Instructions for updating:
Please use `layer.add_weight` method instead.
Model: "sequential_2"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
prune_low_magnitude_dense_2  (None, 20)                822       
_________________________________________________________________
prune_low_magnitude_flatten_ (None, 20)                1         
=================================================================
Total params: 823
Trainable params: 420
Non-trainable params: 403
_________________________________________________________________

Podar algumas camadas (sequencial e funcional)

A poda de um modelo pode ter um efeito negativo na precisão. Você pode podar seletivamente as camadas de um modelo para explorar a compensação entre precisão, velocidade e tamanho do modelo.

Dicas para melhor precisão do modelo:

  • Geralmente é melhor fazer o ajuste fino com a poda, em vez de treinar do zero.
  • Tente podar as camadas posteriores em vez das primeiras.
  • Evite podar camadas críticas (por exemplo, mecanismo de atenção).

Mais :

No exemplo abaixo, podar apenas as camadas Dense .

# Create a base model
base_model = setup_model()
base_model.load_weights(pretrained_weights) # optional but recommended for model accuracy

# Helper function uses `prune_low_magnitude` to make only the 
# Dense layers train with pruning.
def apply_pruning_to_dense(layer):
  if isinstance(layer, tf.keras.layers.Dense):
    return tfmot.sparsity.keras.prune_low_magnitude(layer)
  return layer

# Use `tf.keras.models.clone_model` to apply `apply_pruning_to_dense` 
# to the layers of the model.
model_for_pruning = tf.keras.models.clone_model(
    base_model,
    clone_function=apply_pruning_to_dense,
)

model_for_pruning.summary()
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.iter
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.beta_1
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.beta_2
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.decay
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.learning_rate
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer's state 'm' for (root).layer_with_weights-0.kernel
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer's state 'm' for (root).layer_with_weights-0.bias
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer's state 'v' for (root).layer_with_weights-0.kernel
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer's state 'v' for (root).layer_with_weights-0.bias
WARNING:tensorflow:A checkpoint was restored (e.g. tf.train.Checkpoint.restore or tf.keras.Model.load_weights) but not all checkpointed values were used. See above for specific issues. Use expect_partial() on the load status object, e.g. tf.train.Checkpoint.restore(...).expect_partial(), to silence these warnings, or use assert_consumed() to make the check explicit. See https://www.tensorflow.org/guide/checkpoint#loading_mechanics for details.
Model: "sequential_3"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
prune_low_magnitude_dense_3  (None, 20)                822       
_________________________________________________________________
flatten_3 (Flatten)          (None, 20)                0         
=================================================================
Total params: 822
Trainable params: 420
Non-trainable params: 402
_________________________________________________________________

Embora este exemplo use o tipo de camada para decidir o que podar, a maneira mais fácil de podar uma camada específica é definir sua propriedade de name e procurar esse nome em clone_function .

print(base_model.layers[0].name)
dense_3

Mais legível, mas potencialmente menor precisão do modelo

Isso não é compatível com o ajuste fino com poda, por isso pode ser menos preciso do que os exemplos acima que suportam o ajuste fino.

Embora prune_low_magnitude possa ser aplicado ao definir o modelo inicial, carregar os pesos depois não funciona nos exemplos abaixo.

Exemplo funcional

# Use `prune_low_magnitude` to make the `Dense` layer train with pruning.
i = tf.keras.Input(shape=(20,))
x = tfmot.sparsity.keras.prune_low_magnitude(tf.keras.layers.Dense(10))(i)
o = tf.keras.layers.Flatten()(x)
model_for_pruning = tf.keras.Model(inputs=i, outputs=o)

model_for_pruning.summary()
Model: "functional_1"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
input_1 (InputLayer)         [(None, 20)]              0         
_________________________________________________________________
prune_low_magnitude_dense_4  (None, 10)                412       
_________________________________________________________________
flatten_4 (Flatten)          (None, 10)                0         
=================================================================
Total params: 412
Trainable params: 210
Non-trainable params: 202
_________________________________________________________________

Exemplo sequencial

# Use `prune_low_magnitude` to make the `Dense` layer train with pruning.
model_for_pruning = tf.keras.Sequential([
  tfmot.sparsity.keras.prune_low_magnitude(tf.keras.layers.Dense(20, input_shape=input_shape)),
  tf.keras.layers.Flatten()
])

model_for_pruning.summary()
Model: "sequential_4"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
prune_low_magnitude_dense_5  (None, 20)                822       
_________________________________________________________________
flatten_5 (Flatten)          (None, 20)                0         
=================================================================
Total params: 822
Trainable params: 420
Non-trainable params: 402
_________________________________________________________________

Remova a camada Keras personalizada ou modifique partes da camada para podar

Erro comum: podar o viés geralmente prejudica demais a precisão do modelo.

tfmot.sparsity.keras.PrunableLayer atende a dois casos de uso:

  1. Remova uma camada Keras personalizada
  2. Modifique partes de uma camada interna de Keras para podar.

Por exemplo, o padrão da API é apenas remover o kernel da camada Dense . O exemplo abaixo também corrige o viés.

class MyDenseLayer(tf.keras.layers.Dense, tfmot.sparsity.keras.PrunableLayer):

  def get_prunable_weights(self):
    # Prune bias also, though that usually harms model accuracy too much.
    return [self.kernel, self.bias]

# Use `prune_low_magnitude` to make the `MyDenseLayer` layer train with pruning.
model_for_pruning = tf.keras.Sequential([
  tfmot.sparsity.keras.prune_low_magnitude(MyDenseLayer(20, input_shape=input_shape)),
  tf.keras.layers.Flatten()
])

model_for_pruning.summary()

Model: "sequential_5"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
prune_low_magnitude_my_dense (None, 20)                843       
_________________________________________________________________
flatten_6 (Flatten)          (None, 20)                0         
=================================================================
Total params: 843
Trainable params: 420
Non-trainable params: 423
_________________________________________________________________

Modelo de trem

Model.fit

Chame o retorno de chamada tfmot.sparsity.keras.UpdatePruningStep durante o treinamento.

Para ajudar a depurar o treinamento, use o retorno de chamada tfmot.sparsity.keras.PruningSummaries .

# Define the model.
base_model = setup_model()
base_model.load_weights(pretrained_weights) # optional but recommended for model accuracy
model_for_pruning = tfmot.sparsity.keras.prune_low_magnitude(base_model)

log_dir = tempfile.mkdtemp()
callbacks = [
    tfmot.sparsity.keras.UpdatePruningStep(),
    # Log sparsity and other metrics in Tensorboard.
    tfmot.sparsity.keras.PruningSummaries(log_dir=log_dir)
]

model_for_pruning.compile(
      loss=tf.keras.losses.categorical_crossentropy,
      optimizer='adam',
      metrics=['accuracy']
)

model_for_pruning.fit(
    x_train,
    y_train,
    callbacks=callbacks,
    epochs=2,
)

%tensorboard --logdir={log_dir}
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.iter
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.beta_1
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.beta_2
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.decay
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.learning_rate
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer's state 'm' for (root).layer_with_weights-0.kernel
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer's state 'm' for (root).layer_with_weights-0.bias
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer's state 'v' for (root).layer_with_weights-0.kernel
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer's state 'v' for (root).layer_with_weights-0.bias
WARNING:tensorflow:A checkpoint was restored (e.g. tf.train.Checkpoint.restore or tf.keras.Model.load_weights) but not all checkpointed values were used. See above for specific issues. Use expect_partial() on the load status object, e.g. tf.train.Checkpoint.restore(...).expect_partial(), to silence these warnings, or use assert_consumed() to make the check explicit. See https://www.tensorflow.org/guide/checkpoint#loading_mechanics for details.
Epoch 1/2
1/1 [==============================] - 0s 3ms/step - loss: 1.2485 - accuracy: 0.0000e+00
Epoch 2/2
WARNING:tensorflow:From /tmpfs/src/tf_docs_env/lib/python3.6/site-packages/tensorflow/python/ops/summary_ops_v2.py:1277: stop (from tensorflow.python.eager.profiler) is deprecated and will be removed after 2020-07-01.
Instructions for updating:
use `tf.profiler.experimental.stop` instead.
1/1 [==============================] - 0s 2ms/step - loss: 1.1999 - accuracy: 0.0000e+00

Para não usuários do Colab, você pode ver os resultados de uma execução anterior deste bloco de código em TensorBoard.dev .

Loop de treinamento personalizado

Chame o retorno de chamada tfmot.sparsity.keras.UpdatePruningStep durante o treinamento.

Para ajudar a depurar o treinamento, use o retorno de chamada tfmot.sparsity.keras.PruningSummaries .

# Define the model.
base_model = setup_model()
base_model.load_weights(pretrained_weights) # optional but recommended for model accuracy
model_for_pruning = tfmot.sparsity.keras.prune_low_magnitude(base_model)

# Boilerplate
loss = tf.keras.losses.categorical_crossentropy
optimizer = tf.keras.optimizers.Adam()
log_dir = tempfile.mkdtemp()
unused_arg = -1
epochs = 2
batches = 1 # example is hardcoded so that the number of batches cannot change.

# Non-boilerplate.
model_for_pruning.optimizer = optimizer
step_callback = tfmot.sparsity.keras.UpdatePruningStep()
step_callback.set_model(model_for_pruning)
log_callback = tfmot.sparsity.keras.PruningSummaries(log_dir=log_dir) # Log sparsity and other metrics in Tensorboard.
log_callback.set_model(model_for_pruning)

step_callback.on_train_begin() # run pruning callback
for _ in range(epochs):
  log_callback.on_epoch_begin(epoch=unused_arg) # run pruning callback
  for _ in range(batches):
    step_callback.on_train_batch_begin(batch=unused_arg) # run pruning callback

    with tf.GradientTape() as tape:
      logits = model_for_pruning(x_train, training=True)
      loss_value = loss(y_train, logits)
      grads = tape.gradient(loss_value, model_for_pruning.trainable_variables)
      optimizer.apply_gradients(zip(grads, model_for_pruning.trainable_variables))

  step_callback.on_epoch_end(batch=unused_arg) # run pruning callback

%tensorboard --logdir={log_dir}
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.iter
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.beta_1
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.beta_2
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.decay
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.learning_rate
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer's state 'm' for (root).layer_with_weights-0.kernel
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer's state 'm' for (root).layer_with_weights-0.bias
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer's state 'v' for (root).layer_with_weights-0.kernel
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer's state 'v' for (root).layer_with_weights-0.bias
WARNING:tensorflow:A checkpoint was restored (e.g. tf.train.Checkpoint.restore or tf.keras.Model.load_weights) but not all checkpointed values were used. See above for specific issues. Use expect_partial() on the load status object, e.g. tf.train.Checkpoint.restore(...).expect_partial(), to silence these warnings, or use assert_consumed() to make the check explicit. See https://www.tensorflow.org/guide/checkpoint#loading_mechanics for details.

Para não usuários do Colab, você pode ver os resultados de uma execução anterior deste bloco de código em TensorBoard.dev .

Melhorar a precisão do modelo podado

Primeiro, olhe para a documentação da API tfmot.sparsity.keras.prune_low_magnitude para entender o que é uma programação de poda e a matemática de cada tipo de programação de poda.

Dicas :

  • Tenha uma taxa de aprendizado que não seja muito alta ou muito baixa quando o modelo estiver sendo podado. Considere a programação de poda como um hiperparâmetro.

  • Como um teste rápido, experimente podar um modelo até a dispersão final no início do treinamento, definindo begin_step como 0 com um cronograma tfmot.sparsity.keras.ConstantSparsity . Você pode ter sorte com bons resultados.

  • Não podar com muita frequência para dar ao modelo tempo para se recuperar. A programação de poda fornece uma frequência padrão decente.

  • Para obter ideias gerais para melhorar a precisão do modelo, procure dicas para seu (s) caso (s) de uso em "Definir modelo".

Checkpoint e desserializar

Você deve preservar a etapa do otimizador durante o checkpoint. Isso significa que embora você possa usar os modelos Keras HDF5 para pontos de verificação, não pode usar os pesos Keras HDF5.

# Define the model.
base_model = setup_model()
base_model.load_weights(pretrained_weights) # optional but recommended for model accuracy
model_for_pruning = tfmot.sparsity.keras.prune_low_magnitude(base_model)

_, keras_model_file = tempfile.mkstemp('.h5')

# Checkpoint: saving the optimizer is necessary (include_optimizer=True is the default).
model_for_pruning.save(keras_model_file, include_optimizer=True)
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.iter
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.beta_1
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.beta_2
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.decay
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.learning_rate
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer's state 'm' for (root).layer_with_weights-0.kernel
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer's state 'm' for (root).layer_with_weights-0.bias
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer's state 'v' for (root).layer_with_weights-0.kernel
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer's state 'v' for (root).layer_with_weights-0.bias
WARNING:tensorflow:A checkpoint was restored (e.g. tf.train.Checkpoint.restore or tf.keras.Model.load_weights) but not all checkpointed values were used. See above for specific issues. Use expect_partial() on the load status object, e.g. tf.train.Checkpoint.restore(...).expect_partial(), to silence these warnings, or use assert_consumed() to make the check explicit. See https://www.tensorflow.org/guide/checkpoint#loading_mechanics for details.

O acima se aplica geralmente. O código abaixo é necessário apenas para o formato de modelo HDF5 (não pesos HDF5 e outros formatos).

# Deserialize model.
with tfmot.sparsity.keras.prune_scope():
  loaded_model = tf.keras.models.load_model(keras_model_file)

loaded_model.summary()
WARNING:tensorflow:No training configuration found in the save file, so the model was *not* compiled. Compile it manually.
Model: "sequential_8"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
prune_low_magnitude_dense_8  (None, 20)                822       
_________________________________________________________________
prune_low_magnitude_flatten_ (None, 20)                1         
=================================================================
Total params: 823
Trainable params: 420
Non-trainable params: 403
_________________________________________________________________

Implantar modelo podado

Exportar modelo com compressão de tamanho

Erro comum : tanto strip_pruning quanto a aplicação de um algoritmo de compressão padrão (por exemplo, via gzip) são necessários para ver os benefícios de compressão da poda.

# Define the model.
base_model = setup_model()
base_model.load_weights(pretrained_weights) # optional but recommended for model accuracy
model_for_pruning = tfmot.sparsity.keras.prune_low_magnitude(base_model)

# Typically you train the model here.

model_for_export = tfmot.sparsity.keras.strip_pruning(model_for_pruning)

print("final model")
model_for_export.summary()

print("\n")
print("Size of gzipped pruned model without stripping: %.2f bytes" % (get_gzipped_model_size(model_for_pruning)))
print("Size of gzipped pruned model with stripping: %.2f bytes" % (get_gzipped_model_size(model_for_export)))
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.iter
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.beta_1
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.beta_2
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.decay
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.learning_rate
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer's state 'm' for (root).layer_with_weights-0.kernel
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer's state 'm' for (root).layer_with_weights-0.bias
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer's state 'v' for (root).layer_with_weights-0.kernel
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer's state 'v' for (root).layer_with_weights-0.bias
WARNING:tensorflow:A checkpoint was restored (e.g. tf.train.Checkpoint.restore or tf.keras.Model.load_weights) but not all checkpointed values were used. See above for specific issues. Use expect_partial() on the load status object, e.g. tf.train.Checkpoint.restore(...).expect_partial(), to silence these warnings, or use assert_consumed() to make the check explicit. See https://www.tensorflow.org/guide/checkpoint#loading_mechanics for details.
final model
Model: "sequential_9"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
dense_9 (Dense)              (None, 20)                420       
_________________________________________________________________
flatten_10 (Flatten)         (None, 20)                0         
=================================================================
Total params: 420
Trainable params: 420
Non-trainable params: 0
_________________________________________________________________


Size of gzipped pruned model without stripping: 3299.00 bytes
Size of gzipped pruned model with stripping: 2876.00 bytes

Otimizações específicas de hardware

Uma vez que back-ends diferentes permitem a remoção para melhorar a latência , o uso da dispersão de blocos pode melhorar a latência para determinados hardwares.

Aumentar o tamanho do bloco diminuirá a dispersão de pico que pode ser alcançada para a precisão do modelo de destino. Apesar disso, a latência ainda pode melhorar.

Para obter detalhes sobre o que é compatível com a dispersão de blocos, consulte os documentos da API tfmot.sparsity.keras.prune_low_magnitude

base_model = setup_model()

# For using intrinsics on a CPU with 128-bit registers, together with 8-bit
# quantized weights, a 1x16 block size is nice because the block perfectly
# fits into the register.
pruning_params = {'block_size': [1, 16]}
model_for_pruning = tfmot.sparsity.keras.prune_low_magnitude(base_model, **pruning_params)

model_for_pruning.summary()
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.iter
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.beta_1
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.beta_2
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.decay
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.learning_rate
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer's state 'm' for (root).layer_with_weights-0.kernel
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer's state 'm' for (root).layer_with_weights-0.bias
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer's state 'v' for (root).layer_with_weights-0.kernel
WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer's state 'v' for (root).layer_with_weights-0.bias
WARNING:tensorflow:A checkpoint was restored (e.g. tf.train.Checkpoint.restore or tf.keras.Model.load_weights) but not all checkpointed values were used. See above for specific issues. Use expect_partial() on the load status object, e.g. tf.train.Checkpoint.restore(...).expect_partial(), to silence these warnings, or use assert_consumed() to make the check explicit. See https://www.tensorflow.org/guide/checkpoint#loading_mechanics for details.
Model: "sequential_10"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
prune_low_magnitude_dense_10 (None, 20)                822       
_________________________________________________________________
prune_low_magnitude_flatten_ (None, 20)                1         
=================================================================
Total params: 823
Trainable params: 420
Non-trainable params: 403
_________________________________________________________________