Unable to Load Non-Train Splits
Thank you for preserving this subset of The Pile! For some reason, when I try loading any non-train split, I receive the below error:
Resolving data files: 100%|ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ| 30/30 [00:00<00:00, 237.79it/s]
Downloading data: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ| 30/30 [00:00<00:00, 12537.78files/s]
Generating train split: 177009652 examples [2:29:57, 19672.63 examples/s]
Error: Unknown split "test". Should be one of ['train'].
Can you please advise?
Hey! I had this problem too. I think it's because the train jsonl.zst
files are in a folder, whereas validation and test files aren't.
I did the following to get validation:
pile = load_dataset(
"monology/pile-uncopyrighted",
data_files={"validation": "val.jsonl.zst"},
split="validation",
cache_dir="./mycachedir/"
)
It's seems to be working, I get 179'996 examples. Hope that helps!