python - CNN 模型训练 - 资源耗尽(Python 和 Tensorflow)

Question

我正在使用 Microsoft Azure 训练 CNN(卷积神经网络)，以使用 16k 图像识别 11 类食物。我使用的虚拟机是“STANDARD_NC24_PROMO”，具有以下规范:24 个 vCPU、4 个 GPU、224 GB 内存、1440 GB 存储。

问题是，在简单运行程序时，我收到以下有关资源耗尽的错误:

2-conv-256-nodes-0-dense-1576530179
Train on 10636 samples, validate on 2660 samples
Epoch 1/10
   32/10636 [..............................] - ETA: 57:51
---------------------------------------------------------------------------
ResourceExhaustedError                    Traceback (most recent call last)
<ipython-input-10-ee913a07a18b> in <module>
     86             model.compile(loss="sparse_categorical_crossentropy",optimizer="adam",metrics=["accuracy"])
     87             ### TRAIN
---> 88             model.fit(train_images, train_labels,validation_split=0.20, epochs=10,use_multiprocessing=True)
     89 
     90             loss, acc = model.evaluate(test_images, test_labels, verbose = 0)

/anaconda/envs/azureml_py36/lib/python3.6/site-packages/tensorflow_core/python/keras/engine/training.py in fit(self, x, y, batch_size, epochs, verbose, callbacks, validation_split, validation_data, shuffle, class_weight, sample_weight, initial_epoch, steps_per_epoch, validation_steps, validation_freq, max_queue_size, workers, use_multiprocessing, **kwargs)
    726         max_queue_size=max_queue_size,
    727         workers=workers,
--> 728         use_multiprocessing=use_multiprocessing)
    729 
    730   def evaluate(self,

/anaconda/envs/azureml_py36/lib/python3.6/site-packages/tensorflow_core/python/keras/engine/training_v2.py in fit(self, model, x, y, batch_size, epochs, verbose, callbacks, validation_split, validation_data, shuffle, class_weight, sample_weight, initial_epoch, steps_per_epoch, validation_steps, validation_freq, **kwargs)
    322                 mode=ModeKeys.TRAIN,
    323                 training_context=training_context,
--> 324                 total_epochs=epochs)
    325             cbks.make_logs(model, epoch_logs, training_result, ModeKeys.TRAIN)
    326 

/anaconda/envs/azureml_py36/lib/python3.6/site-packages/tensorflow_core/python/keras/engine/training_v2.py in run_one_epoch(model, iterator, execution_function, dataset_size, batch_size, strategy, steps_per_epoch, num_samples, mode, training_context, total_epochs)
    121         step=step, mode=mode, size=current_batch_size) as batch_logs:
    122       try:
--> 123         batch_outs = execution_function(iterator)
    124       except (StopIteration, errors.OutOfRangeError):
    125         # TODO(kaftan): File bug about tf function and errors.OutOfRangeError?

/anaconda/envs/azureml_py36/lib/python3.6/site-packages/tensorflow_core/python/keras/engine/training_v2_utils.py in execution_function(input_fn)
     84     # `numpy` translates Tensors to values in Eager mode.
     85     return nest.map_structure(_non_none_constant_value,
---> 86                               distributed_function(input_fn))
     87 
     88   return execution_function

/anaconda/envs/azureml_py36/lib/python3.6/site-packages/tensorflow_core/python/eager/def_function.py in __call__(self, *args, **kwds)
    455 
    456     tracing_count = self._get_tracing_count()
--> 457     result = self._call(*args, **kwds)
    458     if tracing_count == self._get_tracing_count():
    459       self._call_counter.called_without_tracing()

/anaconda/envs/azureml_py36/lib/python3.6/site-packages/tensorflow_core/python/eager/def_function.py in _call(self, *args, **kwds)
    518         # Lifting succeeded, so variables are initialized and we can run the
    519         # stateless function.
--> 520         return self._stateless_fn(*args, **kwds)
    521     else:
    522       canon_args, canon_kwds = \

/anaconda/envs/azureml_py36/lib/python3.6/site-packages/tensorflow_core/python/eager/function.py in __call__(self, *args, **kwargs)
   1821     """Calls a graph function specialized to the inputs."""
   1822     graph_function, args, kwargs = self._maybe_define_function(args, kwargs)
-> 1823     return graph_function._filtered_call(args, kwargs)  # pylint: disable=protected-access
   1824 
   1825   @property

/anaconda/envs/azureml_py36/lib/python3.6/site-packages/tensorflow_core/python/eager/function.py in _filtered_call(self, args, kwargs)
   1139          if isinstance(t, (ops.Tensor,
   1140                            resource_variable_ops.BaseResourceVariable))),
-> 1141         self.captured_inputs)
   1142 
   1143   def _call_flat(self, args, captured_inputs, cancellation_manager=None):

/anaconda/envs/azureml_py36/lib/python3.6/site-packages/tensorflow_core/python/eager/function.py in _call_flat(self, args, captured_inputs, cancellation_manager)
   1222     if executing_eagerly:
   1223       flat_outputs = forward_function.call(
-> 1224           ctx, args, cancellation_manager=cancellation_manager)
   1225     else:
   1226       gradient_name = self._delayed_rewrite_functions.register()

/anaconda/envs/azureml_py36/lib/python3.6/site-packages/tensorflow_core/python/eager/function.py in call(self, ctx, args, cancellation_manager)
    509               inputs=args,
    510               attrs=("executor_type", executor_type, "config_proto", config),
--> 511               ctx=ctx)
    512         else:
    513           outputs = execute.execute_with_cancellation(

/anaconda/envs/azureml_py36/lib/python3.6/site-packages/tensorflow_core/python/eager/execute.py in quick_execute(op_name, num_outputs, inputs, attrs, ctx, name)
     65     else:
     66       message = e.message
---> 67     six.raise_from(core._status_to_exception(e.code, message), None)
     68   except TypeError as e:
     69     keras_symbolic_tensors = [

/anaconda/envs/azureml_py36/lib/python3.6/site-packages/six.py in raise_from(value, from_value)

ResourceExhaustedError:  OOM when allocating tensor with shape[32,256,98,98] and type float on /job:localhost/replica:0/task:0/device:GPU:0 by allocator GPU_0_bfc
     [[node sequential_7/conv2d_14/Conv2D (defined at /anaconda/envs/azureml_py36/lib/python3.6/site-packages/tensorflow_core/python/framework/ops.py:1751) ]]
Hint: If you want to see a list of allocated tensors when OOM happens, add report_tensor_allocations_upon_oom to RunOptions for current allocation info.
 [Op:__inference_distributed_function_7727]

Function call stack:
distributed_function

我将在下面附上进行训练的代码:

for dense_layer in dense_layers:
    for layer_size in layer_sizes:
        for conv_layer in conv_layers:
            NAME="{}-conv-{}-nodes-{}-dense-{}".format(conv_layer,
                layer_size, dense_layer, int(time.time()))
            print(NAME)

            model = Sequential()

            model.add(Conv2D(layer_size,(3,3),input_shape=(IMG_SIZE, IMG_SIZE, 1)))
            model.add(Activation("relu"))
            model.add(MaxPooling2D(pool_size=(2,2)))
            model.add(Dropout(0.5))

            for l in range(conv_layer-1):
                model.add(Conv2D(layer_size,(3,3)))
                model.add(Activation("relu"))
                model.add(MaxPooling2D(pool_size=(2,2)))
                model.add(Dropout(0.5))

            model.add(Flatten())
            for l in range(dense_layer):

                model.add(Dense(layer_size))
                model.add(Activation("relu"))

            #The output layer with 11 neurons
            model.add(Dense(11))
            model.add(Activation("softmax"))


            ### COMPILE MODEL
            model.compile(loss="sparse_categorical_crossentropy",
                                            optimizer="adam",
                                            metrics=["accuracy"])
            ### TRAIN
            model.fit(train_images, train_labels,validation_split=0.20, epochs=10)

            loss, acc = model.evaluate(test_images, test_labels, verbose = 0)
            print(acc * 100)
            if maxacc<acc*100:
                maxacc=acc*100
                maxname=NAME
                maxdict[maxacc]=maxname
                print("\n\n",maxacc," ",maxname)

我的笔记本电脑远没有那么好，执行此操作没有问题，但在 azure 上运行它却给了我这个错误。迭代变量并不重要，因为无论它们的值是什么，我仍然会收到错误。

任何帮助将不胜感激，感谢您的宝贵时间!

我想补充一点，该程序甚至无法处理这么少量的层:

dense_layers = [0]
layer_sizes = [32]
conv_layers = [1]

Answer 1

不幸的是，我从未使用 azure 来训练某种网络。但我会尝试:

• 简化您的网络和设置，也许首先使用功能强大的单 GPU。另外，在使用更简单的方法使其工作后，找出必须更改哪些超参数才能使其失败
• 减少批量大小。大多数 GPU OOM 异常都是由于一次处理的数据过多造成的。

发生了很多优化，可能会导致它在本地工作，但对于多 GPU 机器的工作方式略有不同。