Back to Cosmos

ShuffeNet architecture

code/artificial_intelligence/src/shufflenet_v1/README.md

latest844 B
Original Source

ShuffeNet architecture

Implementation of ShuffleNet architecture using CIFAR 10 dataset using Pytorch. The architecture is inspired from the original ShuffleNet paper.

One can replicate the same results by following these steps:

  1. Downloading this jupyter notebook on Google Colab. Alternatively, they can also load the dataset to their own computers instead of using Google Drive (as done in this notebook).
  2. Using GPU instead of CPU in google colab or similarly for your personal computer. In Google Colab, this can be done by going to Edit -> Notebook Settings -> Select GPU -> Save. The training happens faster with GPU as compared to CPU.
  3. Running all the cells sequentially in the order as in the notebook.
  4. It's done! You can go ahead and try the same architecture on different datasets.