matlab-deep-learning / MATLAB-Deep-Learning-Model-Hub

Discover pretrained models for deep learning in MATLAB
https://www.mathworks.com/solutions/deep-learning.html
Other
447 stars 109 forks source link

CUDNN_STAT​US_EXECUTI​ON_FAILED while using convolution2dLayer with 2D image sequence(SSCBT) #7

Open Shipley1105 opened 11 months ago

Shipley1105 commented 11 months ago
lgraph = [...
    sequenceInputLayer([64 5 1], "MinLength",4)
    % convolution2dLayer([3,3],1) % If include this conv2d layer cause
    % ERROR CUDNN_STATUS_EXECUTION_FAILED
    flattenLayer
    gruLayer(16)
    fullyConnectedLayer(16)
    fullyConnectedLayer(2)
    softmaxLayer
    classificationLayer
    ]

If include convolution2dLayer will throw ERROR: Unexpected error calling cuDNN: CUDNN_STATUS_EXECUTION_FAILED

lgraph = [...
    sequenceInputLayer([64 5 1], "MinLength",4)
    sequenceFoldingLayer('Name','fold')
    convolution2dLayer([3,3],1)
    sequenceUnfoldingLayer('Name','unfold')
    flattenLayer
    gruLayer(16)
    fullyConnectedLayer(16)
    fullyConnectedLayer(2)
    softmaxLayer
    classificationLayer
    ]
lgraph = connectLayers(lgraph,'fold/miniBatchSize','unfold/miniBatchSize');

I try to add sequenceFoldingLayer and sequenceUnfoldingLayer, the ERROR still exists. Any help is greatly appreciated.