未验证 提交 82a2f8b7 编写于 作者: L liangym 提交者: GitHub

fix finetune batch size (#2457) (#2460)

* fix batch_size, test=tts
上级 615da211
......@@ -131,10 +131,10 @@ def train_sp(args, config):
converters=converters, )
# collate function and dataloader
train_batch_size = min(len(train_metadata), config.batch_size)
train_sampler = DistributedBatchSampler(
train_dataset,
batch_size=config.batch_size,
batch_size=train_batch_size,
shuffle=True,
drop_last=True)
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册