speech_recognition/FunASR
1
0
Fork 0
mirror of https://github.com/modelscope/FunASR synced 2025-09-15 14:48:36 +08:00
Code Issues Actions Packages Projects Releases Wiki Activity

fix bug

Browse Source
This commit is contained in:
游雁 2024-06-09 03:26:32 +08:00
parent 1186cd96a5
commit 6e0d138826
1 changed files with 7 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

13
funasr/datasets/openai_datasets/datasets.py
View File

@ -211,14 +211,15 @@ class OpenAIDataset(torch.utils.data.Dataset):
if self.batch_type != "example":
b, t = outputs["input_ids"].shape
if b * t > self.batch_size * self.batch_size_scale_ratio_max:
beg = torch.randint(0, 2, ()).item()
if b < 2:
beg = 0
if b > 1 and b * t > self.batch_size * self.batch_size_scale_ratio_max:
# beg = torch.randint(0, 2, ()).item()
# if b < 2:
# beg = 0
logging.info(
f"Warning, b * t: {b * t} > {self.batch_size_scale_ratio_max} * {self.batch_size}, b: {b}, t: {t}, drop half data {idx}th, beg:{beg}"
f"Warning, b*t: {b}*{t}={b * t} > batch_size*relax: {self.batch_size_scale_ratio_max}*{self.batch_size}={self.batch_size_scale_ratio_max*self.batch_size}, drop half data {idx}th, beg:{beg}"
)
samples = samples[beg : beg + b : 2]
# samples = samples[beg : beg + b : 2]
samples = samples[:-1]
continue
break