fix split name

This commit is contained in:
51616 2025-01-05 06:58:54 +00:00
parent 9d22b804b7
commit 831e057eca

View file

@ -389,10 +389,10 @@ def main(output_dir):
"train": tokenized_ds["train"].select(val_train_indices),
}
if "validation" in tokenized_ds:
val_ds["validation"] = tokenized_ds["validation"]
val_ds_size = len(val_ds["validation"])
val_ds["val"] = tokenized_ds["validation"]
val_ds_size = len(val_ds["val"])
val_indices = np.random.permutation(val_ds_size)[: data_args.max_val_samples]
val_ds["validation"] = val_ds["validation"].select(val_indices)
val_ds["val"] = val_ds["val"].select(val_indices)
test_ds = tokenized_ds.get("test", None)
logger.info(f"train_ds: {train_ds}")