80% split
This commit is contained in:
parent
a436a38ddc
commit
dda04a1d5c
@ -34,7 +34,7 @@ def main(config, hug_token):
|
|||||||
|
|
||||||
added_tokens = []
|
added_tokens = []
|
||||||
|
|
||||||
dataset = load_dataset(config.dataset_path)
|
dataset = load_dataset(config.dataset_path, split="train[:80%]")
|
||||||
dataset = dataset.train_test_split(test_size=0.1)
|
dataset = dataset.train_test_split(test_size=0.1)
|
||||||
|
|
||||||
train_dataset = DonutDataset(
|
train_dataset = DonutDataset(
|
||||||
|
Loading…
Reference in New Issue
Block a user