-
Remove README until project is public
-
Remove superseded old-format preprocessed file (replaced by {split}.pt)
-
Remove superseded old-format preprocessed file (replaced by {split}.pt)
-
Remove superseded old-format preprocessed file (replaced by {split}.pt)
-
Re-store train collated split in mmap-friendly format (packed string tensors)
-
Re-store test collated split in mmap-friendly format (packed string tensors)
-
Re-store val collated split in mmap-friendly format (packed string tensors)
-
Add collated preprocessed splits (float32, fast torch.load with weights_only=True)
-
Upload dataset
-
initial commit