PCQM4Mv2 / preprocessed

Commit History

  1. Re-store train collated split in mmap-friendly format (packed string tensors)

    Verified signature
    0f46ae7
  2. Add summary for preprocessed train split

    Verified signature
    5fcd4de
  3. Add collated preprocessed train split (float32, fast torch.load)

    Verified signature
    2744e8f