未验证 提交 ee95ba00 编写于 作者: G Guanghua Yu 提交者: GitHub

fix error in train_batch_size sync from load_config (#1568)

上级 84dde59e
......@@ -139,7 +139,7 @@ def merge_config(config, another_cfg=None):
# batch size config to global, models can get batch size config
# from global config when building model.
# batch size in evaluation or inference can also be added here
if 'TrainReader' in dct:
if 'TrainReader' in dct and 'batch_size' in dct['TrainReader']:
dct['train_batch_size'] = dct['TrainReader']['batch_size']
return dct
......
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