The Formidable Repository of Open Sparse Tensors and Tools

Home » Contribute » Contributing to FROSTT

Contributing to FROSTT

FROSTT is an open source project that thrives off community contributions.

FROSTT Development

Want to contribute to the development of this site or its tools? Great! All development is conducted on Github. Feel free to file an issue or submit a pull request.

Tensor Construction

Sparse tensors are constructed from a large variety of data sources. These sources are often large, unstructured, and noisy. To aid in the construction of tensors from these data sources, FROSTT has released a tool for parsing CSV-like data. If your data can be represented as a CSV file, our tensor_parser tool can map features of data (represented as columns) to tensor modes and handle many of the tedious details such as mapping entities to contiguous indices, sorting data, and merging duplicate non-zeros,

tensor_parser is MIT licensed and can be found on GitHub.

Tensor Submission

If you have a tensor that you would like to see in this collection, please consider sharing with us. You must have the rights to distribute the data. By sharing your dataset, you acknowledge that the data will become publicly available.

  1. Fork on Github.

  2. Copy the provided template to the _tensors/ directory. The name of the file will form the URL of the tensor page. Suppose your new tensor is named Big-Tensor: $ cp _tensors/

  3. Fill in as much information as possible about your new tensor.

  4. Build your tensor page. If all went well, your tensor should now exist. Reload your local FROSTT page and Big-Tensor should now be available.

  5. Once the tensor page is ready, submit a pull request and include a public link to the tensor data. We will copy the data to our hosting location, update the links to our own hosting, merge the pull request, and go live!