Skip to content

Track 2 (ML): Models similar to the Cerebras modelzoo models

Please, following the model porting instruction in Port Pytorch Models To Cerebras Documentation to learn about ow to convert existing Pytorch models to run on Cerebras systems.

Track Specific Questions

If your project falls under this category, make sure to address the following questions in your application document:

  1. Please, attach a diagram that describes your proposed model, along the lines illustrated in this example: Figure: The Transformer, model architecture (Vaswani A. et. al. 2017).
  2. How big is the model you intend to train on Neocortex? Please, describe in terms of the number of layers or kernels that comprise it. Include size in terms of MB as well.
  3. How big is the dataset (number of samples, MB per sample) you intend to use?
  4. Please specify the shapes of the input and output tensors for your model/s.
  5. If possible, please specify the name of the dimensions for your input and output tensors from the previous question. E.g. (batch, input channels, height, width)
  6. Please specify the loss function that you would like to use.
  7. Please, list the libraries complementary to standard PyTorch and/or TensorFlow distributions that you would need to train your model(s).
  8. Please, list the key libraries that you would need for data preprocessing.