CNTK out of memory error when model.fit() is called second time

Question

I am using Keras and CNTK(backend)

my code is like this:

def run_han(embeddings_index, fname, opt)
    ...
    sentence_input = Input(shape=(MAX_SENT_LENGTH,), dtype='int32')
    embedded_sequences = embedding_layer(sentence_input)
    l_lstm = Bidirectional(GRU(GRU_UNITS, return_sequences=True, kernel_regularizer=l2_reg, 
                           implementation=GPU_IMPL))(embedded_sequences)
    l_att = AttLayer(regularizer=l2_reg)(l_lstm)            
    sentEncoder = Model(sentence_input, l_att)

    review_input = Input(shape=(MAX_SENTS, MAX_SENT_LENGTH), dtype='int32')
    review_encoder = TimeDistributed(sentEncoder)(review_input)
    l_lstm_sent = Bidirectional(GRU(GRU_UNITS, return_sequences=True, kernel_regularizer=l2_reg, 
                                implementation=GPU_IMPL))(review_encoder)
    l_att_sent = AttLayer(regularizer=l2_reg)(l_lstm_sent) 
    preds = Dense(n_classes, activation='softmax', kernel_regularizer=l2_reg)(l_att_sent)
    model = Model(review_input, preds)

    model.compile(loss='categorical_crossentropy',
          optimizer=opt, #SGD(lr=0.1, nesterov=True),
          metrics=['acc'])
   ...
   model.fit(x_train[ind,:,:], y_train[ind,:], epochs=NUM_EPOCHS, batch_size=BATCH_SIZE, shuffle=False, 
          callbacks=[cr_result, history, csv_logger], 
          verbose=2,validation_data=(x_test, y_test), class_weight = class_weight)
   ...
    %xdel model
    gc.collect()

I call the above model several times as I change optimizer. like this:

opt = optimizers.RMSprop(lr=0.0001, rho=0.9, epsilon=1e-08, decay=0.0, clipvalue=0.5)
run_han(embeddings_index, 'w2v_100_all_rms_cw', opt, class_weight)

opt = optimizers.Adadelta(lr=1.0, rho=0.95, epsilon=1e-08, decay=0.0, clipvalue=0.5)
run_han(embeddings_index, 'w2v_100_all_adadelta_cw', opt, class_weight)

opt =  optimizers.Adagrad(lr=0.01, epsilon=1e-08, decay=0.0, clipvalue=0.5)
run_han(embeddings_index, 'w2v_100_all_adagrad_cw', opt, class_weight)

When the model.fit() is called second time, out of memory error is showing

 RuntimeError: CUDA failure 2: out of memory ; GPU=0 ; hostname=USER-PC ; expr=cudaMalloc((void**) &deviceBufferPtr, sizeof(AllocatedElemType) * AsMultipleOf(numElements, 2))
[CALL STACK]
> Microsoft::MSR::CNTK::CudaTimer::  Stop
- Microsoft::MSR::CNTK::CudaTimer::  Stop (x2)
- Microsoft::MSR::CNTK::GPUMatrix<float>::  Resize
- Microsoft::MSR::CNTK::Matrix<float>::  Resize
- Microsoft::MSR::CNTK::DataTransferer::  operator= (x4)
- CNTK::Internal::  UseSparseGradientAggregationInDataParallelSGD
- Microsoft::MSR::CNTK::DataTransferer::  operator=
- CNTK::Internal::  UseSparseGradientAggregationInDataParallelSGD
- CNTK::Function::  Forward
- CNTK::  CreateTrainer
- CNTK::Trainer::  TotalNumberOfSamplesSeen
- CNTK::Trainer::  TrainMinibatch

I thought it is because the memory of the first run was not released from gpu, So I added this after model.fit()

%xdel model

gc.collect()

However, the error is same. I cannot figure out the cause of error. Is it because of my Keras code or CNTK?

(GTX 1080ti, Window 7, Python 2.7, CNTK 2.2, Jupyter)

Answer 1

This is a really annoying problem and it arises from the fact that for some reason a code compiled to be executed on CPU is not garbage-collected properly. So even though you are running a garbage collector - a compiled model is still on GPU. In order to overcome this, you may try a solution presented here (TLDR: run training in a separate process - as when process is finished - memory is cleared)