Running with GPU

[floatlazer]

Hi,

Thanks for sharing your great work ! I encountered the following error when trying to run your code with GPU in MATLAB R2018a :

Error using vl_myrelu (line 29)
The following error occurred converting from gpuArray to double:
Conversion to double from gpuArray is not possible.

Error in vl_myforbackward (line 110)
        res(i).dzdx = vl_myrelu(res(i), res(i+1).dzdx) ;

Error in lienet_train_g3d>process_epoch (line 122)
    res = vl_myforbackward(net, lie_data, dzdy, res) ;

Error in lienet_train_g3d (line 29)
    [net,stats.train] = process_epoch(opts, epoch, lie_train, train, learningRate, net) ;

Error in lienet_g3d (line 19)
[net, info] = lienet_train_g3d(net, lie_train, opts);

The code runs perfectly on CPUs. Do you have any advice ? Thanks !

Xuan

[hjf1997]

Hi,

Thanks for sharing your great work ! I encountered the following error when trying to run your code with GPU in MATLAB R2018a :

Error using vl_myrelu (line 29)
The following error occurred converting from gpuArray to double:
Conversion to double from gpuArray is not possible.

Error in vl_myforbackward (line 110)
        res(i).dzdx = vl_myrelu(res(i), res(i+1).dzdx) ;

Error in lienet_train_g3d>process_epoch (line 122)
    res = vl_myforbackward(net, lie_data, dzdy, res) ;

Error in lienet_train_g3d (line 29)
    [net,stats.train] = process_epoch(opts, epoch, lie_train, train, learningRate, net) ;

Error in lienet_g3d (line 19)
[net, info] = lienet_train_g3d(net, lie_train, opts);

The code runs perfectly on CPUs. Do you have any advice ? Thanks !

Xuan

I have implemented the pytorch version of this paper, so you could use this code to train on GPU or GPUs~