Awesome! Glad it worked. Would you be able to share your final model so I can open a Github Issue for the performance issue you found? I see that most of the code is here but that has some errors in that you’ve fixed. Would be great to get the fixed up version, and the shape of the input you’re using for these models. Cheers, Thom
Related Topics
Topic | Replies | Views | Activity | |
---|---|---|---|---|
Worse performance of GPU when evaluating MobileNet | 0 | 491 | November 9, 2021 | |
Single-machine multi-GPU training, time is not speeding up | 5 | 2154 | November 16, 2018 | |
MXNet vs Pytorch Benchmark | 3 | 2246 | May 27, 2019 | |
Mxnet with cudnn7.1 is little slower than cudnn5.1 | 1 | 407 | July 13, 2018 | |
`MXImperativeInvokeEx` is taking a long time | 8 | 770 | January 6, 2019 |