hpe-cct / cct-tutorial

Learn how to use CCT!
Apache License 2.0
13 stars 3 forks source link

Adding movie convolution sample, Trials of AlexNet, OverFeat, Vgg_A a… #4

Closed shrutiramesh88 closed 8 years ago

shrutiramesh88 commented 8 years ago

Adding movie convolution sample, Trials of AlexNet, OverFeat, Vgg_A and CIFAR10

mneary1 commented 8 years ago

there are a few paths to data in this commit that are specific to your machine, those should probably be generalized

bchandle commented 8 years ago

Can you break this up and submit it in smaller chunks?

All of the neural network functionality should go to the cct-nn repository instead. The examples in the tutorial repository are a snapshot from the last release of the NN package.

For the convolution example, make sure you have Git LFS installed first (https://git-lfs.github.com). The courtyard video is already in the repository. It's just only visible if you're using Github Desktop or have installed the Git LFS plugin. I probably can't merge any of the other video files. HPE doesn't hold copyright over any of those, so we can't relicense them to Apache without permission.

shrutiramesh88 commented 8 years ago

The code was not meant to be merged, since it is not in the final form. As discussed in the telecom the other day, we just wanted to point to our trial code for AlexNet, OverFeat, Vgg_A and Cifar10. We are facing memory issues on our local server (CL_MEM_OBJECT_ALLOCATION_FAILURE) for larger inputs of AlexNet, OverFeat and Vgg_A. It would be good if you could try our code on one of your machines there, so that we can confirm that it is not the code problem.