我有同样的问题但是当我尝试应用相同的修复时，我遇到了另一个错误。然而，我正在运行5 gpus。我已经读过你需要确保你的样品可以被批次和gpus的数量整除，但我已经这样做了。我已经在网上搜了好几天了，我找不到任何能够解决我遇到的问题的东西。我正在运行keras v2.0.9和张量流v1.1.0
变量： attributeTables [0]是一个numpy数组形状（35560,700） y是一个numpy数组形状（35560，）我也尝试使用shape（35560,1）为y但是所有发生的事情是“不兼容的形状：[2540]与[508]”从那里变为“不兼容的形状： [2540,1]与[508,1]“
所以这告诉我，问题只出在目标上，并且预期的批量大小在过程中间的某处仅为目标而不是导致不匹配的属性或者至少只是在进行验证时增加我不确定。
以下是有问题的代码和错误。
import numpy as np
from keras.models import Sequential
from keras.utils import multi_gpu_model
from keras.layers import Dense
from keras.layers import LSTM
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
import matplotlib.pyplot as plt
def baseline_model():
# create model
print("Building Layers")
model = Sequential()
model.add(LSTM(700, batch_input_shape=(batchSize, X.shape[1], X.shape[2]), activation='tanh', return_sequences=False, stateful=True))
model.add(Dense(1))
print("Building Parallel model")
parallel_model = multi_gpu_model(model, gpus=nGPU)
# Compile model
#model.compile(loss='mean_squared_error', optimizer='adam')
print("Compiling Model")
parallel_model.compile(loss='mae', optimizer='adam', metrics=['accuracy'])
return parallel_model
def buildModel():
print("Bulding Model")
mlp = baseline_model()
print("Fitting Model")
return mlp.fit(X_train, y_train, epochs=1, batch_size=batchSize, shuffle=False, validation_data=(X_test, y_test))
print("Scaling")
scaler = StandardScaler()
X_Scaled = scaler.fit_transform(attributeTables[0])
print("Finding Batch Size")
nGPU = 5
batchSize = 500
while len(X_Scaled) % (batchSize * nGPU) != 0:
batchSize += 1
print("Filling Arrays")
X = X_Scaled.reshape((X_Scaled.shape[0], X_Scaled.shape[1], 1))
X_train, X_test, y_train, y_test = train_test_split(X, y, train_size=.8)
print("Calling buildModel()")
model = buildModel()
print("Ploting History")
plt.plot(model.history['loss'], label='train')
plt.plot(model.history['val_loss'], label='test')
plt.legend()
plt.show()
这是我的完整输出。
Beginning OHLC Load
Time took : 7.571000099182129
Making gloabal copies
Time took : 0.0
Using TensorFlow backend.
Scaling
Finding Batch Size
Filling Arrays
Calling buildModel()
Bulding Model
Building Layers
C:\ProgramData\Anaconda3\lib\site-packages\sklearn\model_selection\_split.py:2010: FutureWarning: From version 0.21, test_size will always complement train_size unless both are specified.
FutureWarning)
Building Parallel model
Compiling Model
Fitting Model
Train on 28448 samples, validate on 7112 samples
Epoch 1/1
Traceback (most recent call last):
File "<ipython-input-2-74c49f05bfbc>", line 1, in <module>
runfile('C:/Users/BeeAndTurtle/Documents/Programming/Python/Kraken_API_Market_Prediction/predictor/test.py', wdir='C:/Users/BeeAndTurtle/Documents/Programming/Python/Kraken_API_Market_Prediction/predictor')
File "C:\ProgramData\Anaconda3\lib\site-packages\spyder\utils\site\sitecustomize.py", line 710, in runfile
execfile(filename, namespace)
File "C:\ProgramData\Anaconda3\lib\site-packages\spyder\utils\site\sitecustomize.py", line 101, in execfile
exec(compile(f.read(), filename, 'exec'), namespace)
File "C:/Users/BeeAndTurtle/Documents/Programming/Python/Kraken_API_Market_Prediction/predictor/test.py", line 77, in <module>
model = buildModel()
File "C:/Users/BeeAndTurtle/Documents/Programming/Python/Kraken_API_Market_Prediction/predictor/test.py", line 57, in buildModel
return mlp.fit(X_train, y_train, epochs=1, batch_size=batchSize, shuffle=False, validation_data=(X_test, y_test))
File "C:\ProgramData\Anaconda3\lib\site-packages\keras\engine\training.py", line 1631, in fit
validation_steps=validation_steps)
File "C:\ProgramData\Anaconda3\lib\site-packages\keras\engine\training.py", line 1213, in _fit_loop
outs = f(ins_batch)
File "C:\ProgramData\Anaconda3\lib\site-packages\keras\backend\tensorflow_backend.py", line 2332, in __call__
**self.session_kwargs)
File "C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\python\client\session.py", line 778, in run
run_metadata_ptr)
File "C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\python\client\session.py", line 982, in _run
feed_dict_string, options, run_metadata)
File "C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\python\client\session.py", line 1032, in _do_run
target_list, options, run_metadata)
File "C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\python\client\session.py", line 1052, in _do_call
raise type(e)(node_def, op, message)
InvalidArgumentError: Incompatible shapes: [2540,1] vs. [508,1]
[[Node: training/Adam/gradients/loss/concatenate_1_loss/sub_grad/BroadcastGradientArgs = BroadcastGradientArgs[T=DT_INT32, _class=["loc:@loss/concatenate_1_loss/sub"], _device="/job:localhost/replica:0/task:0/gpu:0"](training/Adam/gradients/loss/concatenate_1_loss/sub_grad/Shape, training/Adam/gradients/loss/concatenate_1_loss/sub_grad/Shape_1)]]
[[Node: replica_1/sequential_1/dense_1/BiasAdd/_313 = _Recv[client_terminated=false, recv_device="/job:localhost/replica:0/task:0/cpu:0", send_device="/job:localhost/replica:0/task:0/gpu:1", send_device_incarnation=1, tensor_name="edge_1355_replica_1/sequential_1/dense_1/BiasAdd", tensor_type=DT_FLOAT, _device="/job:localhost/replica:0/task:0/cpu:0"]()]]
Caused by op 'training/Adam/gradients/loss/concatenate_1_loss/sub_grad/BroadcastGradientArgs', defined at:
File "C:\ProgramData\Anaconda3\lib\site-packages\spyder\utils\ipython\start_kernel.py", line 245, in <module>
main()
File "C:\ProgramData\Anaconda3\lib\site-packages\spyder\utils\ipython\start_kernel.py", line 241, in main
kernel.start()
File "C:\ProgramData\Anaconda3\lib\site-packages\ipykernel\kernelapp.py", line 477, in start
ioloop.IOLoop.instance().start()
File "C:\ProgramData\Anaconda3\lib\site-packages\zmq\eventloop\ioloop.py", line 177, in start
super(ZMQIOLoop, self).start()
File "C:\ProgramData\Anaconda3\lib\site-packages\tornado\ioloop.py", line 832, in start
self._run_callback(self._callbacks.popleft())
File "C:\ProgramData\Anaconda3\lib\site-packages\tornado\ioloop.py", line 605, in _run_callback
ret = callback()
File "C:\ProgramData\Anaconda3\lib\site-packages\tornado\stack_context.py", line 277, in null_wrapper
return fn(*args, **kwargs)
File "C:\ProgramData\Anaconda3\lib\site-packages\ipykernel\kernelbase.py", line 265, in enter_eventloop
self.eventloop(self)
File "C:\ProgramData\Anaconda3\lib\site-packages\ipykernel\eventloops.py", line 106, in loop_qt5
return loop_qt4(kernel)
File "C:\ProgramData\Anaconda3\lib\site-packages\ipykernel\eventloops.py", line 99, in loop_qt4
_loop_qt(kernel.app)
File "C:\ProgramData\Anaconda3\lib\site-packages\ipykernel\eventloops.py", line 83, in _loop_qt
app.exec_()
File "C:\ProgramData\Anaconda3\lib\site-packages\ipykernel\eventloops.py", line 39, in process_stream_events
kernel.do_one_iteration()
File "C:\ProgramData\Anaconda3\lib\site-packages\ipykernel\kernelbase.py", line 298, in do_one_iteration
stream.flush(zmq.POLLIN, 1)
File "C:\ProgramData\Anaconda3\lib\site-packages\zmq\eventloop\zmqstream.py", line 352, in flush
self._handle_recv()
File "C:\ProgramData\Anaconda3\lib\site-packages\zmq\eventloop\zmqstream.py", line 472, in _handle_recv
self._run_callback(callback, msg)
File "C:\ProgramData\Anaconda3\lib\site-packages\zmq\eventloop\zmqstream.py", line 414, in _run_callback
callback(*args, **kwargs)
File "C:\ProgramData\Anaconda3\lib\site-packages\tornado\stack_context.py", line 277, in null_wrapper
return fn(*args, **kwargs)
File "C:\ProgramData\Anaconda3\lib\site-packages\ipykernel\kernelbase.py", line 283, in dispatcher
return self.dispatch_shell(stream, msg)
File "C:\ProgramData\Anaconda3\lib\site-packages\ipykernel\kernelbase.py", line 235, in dispatch_shell
handler(stream, idents, msg)
File "C:\ProgramData\Anaconda3\lib\site-packages\ipykernel\kernelbase.py", line 399, in execute_request
user_expressions, allow_stdin)
File "C:\ProgramData\Anaconda3\lib\site-packages\ipykernel\ipkernel.py", line 196, in do_execute
res = shell.run_cell(code, store_history=store_history, silent=silent)
File "C:\ProgramData\Anaconda3\lib\site-packages\ipykernel\zmqshell.py", line 533, in run_cell
return super(ZMQInteractiveShell, self).run_cell(*args, **kwargs)
File "C:\ProgramData\Anaconda3\lib\site-packages\IPython\core\interactiveshell.py", line 2698, in run_cell
interactivity=interactivity, compiler=compiler, result=result)
File "C:\ProgramData\Anaconda3\lib\site-packages\IPython\core\interactiveshell.py", line 2808, in run_ast_nodes
if self.run_code(code, result):
File "C:\ProgramData\Anaconda3\lib\site-packages\IPython\core\interactiveshell.py", line 2862, in run_code
exec(code_obj, self.user_global_ns, self.user_ns)
File "<ipython-input-2-74c49f05bfbc>", line 1, in <module>
runfile('C:/Users/BeeAndTurtle/Documents/Programming/Python/Kraken_API_Market_Prediction/predictor/test.py', wdir='C:/Users/BeeAndTurtle/Documents/Programming/Python/Kraken_API_Market_Prediction/predictor')
File "C:\ProgramData\Anaconda3\lib\site-packages\spyder\utils\site\sitecustomize.py", line 710, in runfile
execfile(filename, namespace)
File "C:\ProgramData\Anaconda3\lib\site-packages\spyder\utils\site\sitecustomize.py", line 101, in execfile
exec(compile(f.read(), filename, 'exec'), namespace)
File "C:/Users/BeeAndTurtle/Documents/Programming/Python/Kraken_API_Market_Prediction/predictor/test.py", line 77, in <module>
model = buildModel()
File "C:/Users/BeeAndTurtle/Documents/Programming/Python/Kraken_API_Market_Prediction/predictor/test.py", line 57, in buildModel
return mlp.fit(X_train, y_train, epochs=1, batch_size=batchSize, shuffle=False, validation_data=(X_test, y_test))
File "C:\ProgramData\Anaconda3\lib\site-packages\keras\engine\training.py", line 1608, in fit
self._make_train_function()
File "C:\ProgramData\Anaconda3\lib\site-packages\keras\engine\training.py", line 990, in _make_train_function
loss=self.total_loss)
File "C:\ProgramData\Anaconda3\lib\site-packages\keras\legacy\interfaces.py", line 87, in wrapper
return func(*args, **kwargs)
File "C:\ProgramData\Anaconda3\lib\site-packages\keras\optimizers.py", line 415, in get_updates
grads = self.get_gradients(loss, params)
File "C:\ProgramData\Anaconda3\lib\site-packages\keras\optimizers.py", line 73, in get_gradients
grads = K.gradients(loss, params)
File "C:\ProgramData\Anaconda3\lib\site-packages\keras\backend\tensorflow_backend.py", line 2369, in gradients
return tf.gradients(loss, variables, colocate_gradients_with_ops=True)
File "C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\python\ops\gradients_impl.py", line 560, in gradients
grad_scope, op, func_call, lambda: grad_fn(op, *out_grads))
File "C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\python\ops\gradients_impl.py", line 368, in _MaybeCompile
return grad_fn() # Exit early
File "C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\python\ops\gradients_impl.py", line 560, in <lambda>
grad_scope, op, func_call, lambda: grad_fn(op, *out_grads))
File "C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\python\ops\math_grad.py", line 609, in _SubGrad
rx, ry = gen_array_ops._broadcast_gradient_args(sx, sy)
File "C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\python\ops\gen_array_ops.py", line 411, in _broadcast_gradient_args
name=name)
File "C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\python\framework\op_def_library.py", line 768, in apply_op
op_def=op_def)
File "C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\python\framework\ops.py", line 2336, in create_op
original_op=self._default_original_op, op_def=op_def)
File "C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\python\framework\ops.py", line 1228, in __init__
self._traceback = _extract_stack()
...which was originally created as op 'loss/concatenate_1_loss/sub', defined at:
File "C:\ProgramData\Anaconda3\lib\site-packages\spyder\utils\ipython\start_kernel.py", line 245, in <module>
main()
[elided 27 identical lines from previous traceback]
File "C:/Users/BeeAndTurtle/Documents/Programming/Python/Kraken_API_Market_Prediction/predictor/test.py", line 77, in <module>
model = buildModel()
File "C:/Users/BeeAndTurtle/Documents/Programming/Python/Kraken_API_Market_Prediction/predictor/test.py", line 55, in buildModel
mlp = baseline_model()
File "C:/Users/BeeAndTurtle/Documents/Programming/Python/Kraken_API_Market_Prediction/predictor/test.py", line 29, in baseline_model
parallel_model.compile(loss='mae', optimizer='adam', metrics=['accuracy'])
File "C:\ProgramData\Anaconda3\lib\site-packages\keras\engine\training.py", line 860, in compile
sample_weight, mask)
File "C:\ProgramData\Anaconda3\lib\site-packages\keras\engine\training.py", line 460, in weighted
score_array = fn(y_true, y_pred)
File "C:\ProgramData\Anaconda3\lib\site-packages\keras\losses.py", line 13, in mean_absolute_error
return K.mean(K.abs(y_pred - y_true), axis=-1)
File "C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\python\ops\math_ops.py", line 821, in binary_op_wrapper
return func(x, y, name=name)
File "C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\python\ops\gen_math_ops.py", line 2627, in _sub
result = _op_def_lib.apply_op("Sub", x=x, y=y, name=name)
File "C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\python\framework\op_def_library.py", line 768, in apply_op
op_def=op_def)
File "C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\python\framework\ops.py", line 2336, in create_op
original_op=self._default_original_op, op_def=op_def)
File "C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\python\framework\ops.py", line 1228, in __init__
self._traceback = _extract_stack()
InvalidArgumentError (see above for traceback): Incompatible shapes: [2540,1] vs. [508,1]
[[Node: training/Adam/gradients/loss/concatenate_1_loss/sub_grad/BroadcastGradientArgs = BroadcastGradientArgs[T=DT_INT32, _class=["loc:@loss/concatenate_1_loss/sub"], _device="/job:localhost/replica:0/task:0/gpu:0"](training/Adam/gradients/loss/concatenate_1_loss/sub_grad/Shape, training/Adam/gradients/loss/concatenate_1_loss/sub_grad/Shape_1)]]
[[Node: replica_1/sequential_1/dense_1/BiasAdd/_313 = _Recv[client_terminated=false, recv_device="/job:localhost/replica:0/task:0/cpu:0", send_device="/job:localhost/replica:0/task:0/gpu:1", send_device_incarnation=1, tensor_name="edge_1355_replica_1/sequential_1/dense_1/BiasAdd", tensor_type=DT_FLOAT, _device="/job:localhost/replica:0/task:0/cpu:0"]()]]
答案 0 :(得分：1)
Daniel Moller的链接是正确的，当我禁用并行模型并将其放在一个GPU上时，有状态的工作没有ptoblem。目前正在等待训练。将发布结果。
答案 1 :(得分：0)
我刚刚发布了一个实验性实用程序stateful_multi_gpu，用于处理多个GPU的状态模型培训。我很想知道它是否对你有用。
请参阅my answer，了解DanielMöller提到的同一问题。