Datasets:
Tasks:
Text Generation
Modalities:
Text
Formats:
parquet
Languages:
code
Size:
10K - 100K
ArXiv:
DOI:
License:
Dataset-specific Folders
#3
by
ludgerpaehler
- opened
With the move to a data
folder, we are able to permit further separation between source languages by e.g. moving Rust out into its own folders, and then only loading the IR generated from R if so desired with e.g.
ds = load_dataset("llvm-ml/ComPile", data_dir="data/rust", split="train")
although it has to be considered how deduplication affects this.
This is done now.
boomanaiden154
changed discussion status to
closed