InternalError Blas GEMM launch failed a shape 128 122 b shape 122 64 m 128 n 64 k 122

Question

When I run this model using python:

def build_network(): models = [] model = Sequential() model.add(Dense(64, input_dim=122)) model.add(Dense(32)) model.add(Activation('relu')) model.add(Dropout(.15)) model.add(Dense(32)) model.add(Activation('relu')) model.add(Dropout(.15)) model.add(Dense(32)) model.add(Activation('relu')) model.add(Dropout(.15)) model.add(Dense(5)) model.add(Activation('softmax')) model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy']) return model NN = build_network() early_stopping = EarlyStopping(monitor='val_loss', min_delta=0, patience=3, verbose=0, mode='auto') NN.fit(x=train, y=target, epochs=100, validation_split=0.1, batch_size=128, callbacks=[early_stopping])

I receive this error:

InternalError: Blas GEMM launch failed : a.shape=(128, 122), b.shape=(122, 64), m=128, n=64, k=122 [[{{node dense_6/MatMul}}]]

Comments

Hey, @Arezoo,

This error is a Cuda error because the GPU is running out of memory, doesn't have enough compute capacity, or there's a driver issue.

Can you confirm that none of these issues is happening with you here?

Answer 1

Hey,@Gitika,

I don't know, could you please help me to check and resolved this problem?

Comments

Hey, @Arezoo,

Do this and revert me back with the output. run nvidia-smi to check you have no other processes using the GPU running and revert me the output you are getting while running the command.

---

Hey,@Gitika,

I run this command in command prompt and received this error,

C:\Users\Administrator> nvidia-smi
'nvidia-smi' is not recognized as an internal or external command,
operable program or batch file.

and I run in the consle of spyder and received this error,

nvidia-smi
Traceback (most recent call last):

  File "<ipython-input-5-a5d968248d17>", line 1, in <module>
    nvidia-smi

NameError: name 'nvidia' is not defined

where do I  execute this command(nvidia-smi)?