Add dataset card, link to paper and code

#1
by nielsr HF Staff - opened

This PR adds a dataset card for the pre-tokenized data used in the paper "Data-Constrained Language Model Pretraining: Improved Regularization and Scaling Laws". It includes:

  • Relevant metadata (task category).
  • Links to the paper and the official GitHub repository.
  • A description of the data format (packed GPT-2 tokens).
  • Sample usage for downloading the dataset using the Hugging Face CLI.
zhiwei555 changed pull request status to merged

Sign up or log in to comment