|
|
@@ -319,7 +319,7 @@ class AutoAugTextDataset(IterableDataset):
|
|
|
|
|
|
all_tokens, all_labels = [], []
|
|
|
while remaining_tokens > 0 and len(samples) > 0:
|
|
|
- sentence = samples.pop()
|
|
|
+ sentence = samples.pop(0)
|
|
|
|
|
|
text = random.choice(sentence.texts)
|
|
|
text, length = self.tokenize_sentence(text)
|
|
|
@@ -648,7 +648,7 @@ if __name__ == "__main__":
|
|
|
from tqdm import tqdm
|
|
|
|
|
|
ds = AutoAugTextDataset(
|
|
|
- ["data/protos/test"],
|
|
|
+ ["data/protos"],
|
|
|
tokenizer=AutoTokenizer.from_pretrained("fishaudio/fish-speech-1"),
|
|
|
use_speaker=False,
|
|
|
interactive_prob=1.0,
|
Lengyue