Deep learning toolbox: Failed to plot animated loss curve during the training process.

1 次查看（过去 30 天）
qi lu
qi lu 2020-12-4
回答： Monisha Nalluru 2020-12-7
I use deep learning toolbox to solve a regression problem. I set the 'Plots' in training options to 'training-progress'. In the intial stage, it plots well. But after several iterations, animated loss curve doesn't update. And the training process doesn't stop. I did not encounter this problem in the Matlab2020a. The present version is matlab 2020b.
% modify the architecture of resnet50 to solve a regression problem.
lgraph = resnet50_regression4(resnet50)
options = trainingOptions('adam', ...
'MiniBatchSize',miniBatchSize, ...
'MaxEpochs',300, ...
'InitialLearnRate',1e-3, ...
'LearnRateSchedule','piecewise', ...
'LearnRateDropPeriod',30, ...
'LearnRateDropFactor',0.2, ...
'Shuffle','every-epoch', ...
'ValidationData',{X_val,Y_val}, ...
'ValidationFrequency',validationFrequency, ...
'ValidationPatience',35, ...
'L2Regularization', 0.0001, ...
'GradientThreshold', 1, ...
'Verbose',true, ...
'VerboseFrequency', validationFrequency, ...
'ExecutionEnvironment', 'gpu', ...
'Plots', 'training-progress', ...
'CheckpointPath', check_path, ...
'OutputFcn',@(info)userplots(info));
[net,Info] = trainNetwork(X_train,Y_train,lgraph,options);

请先登录，再回答此问题。

回答（1 个）

Monisha Nalluru
Monisha Nalluru 2020-12-7
The issue could be caused by Graphics Drivers
Launching MATLAB from command prompt using "-softwareopenglmesa" flag will fix the above issue

类别

在 Help Center 和 File Exchange 中查找有关 Box Plots 的更多信息

标签

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by