提交 2b2e0b59 编写于 作者: H Hongkun Yu 提交者: A. Unique TensorFlower

The tf_data_service override only works for the TaskConfig.

PiperOrigin-RevId: 339488247
上级 290cd97b
......@@ -82,15 +82,19 @@ def parse_configuration(flags_obj):
'runtime': {
'tpu': flags_obj.tpu,
},
'task': {
'train_data': {
'tf_data_service_address': flags_obj.tf_data_service,
},
'validation_data': {
'tf_data_service_address': flags_obj.tf_data_service,
}
}
})
if flags_obj.tf_data_service and isinstance(params.task,
config_definitions.TaskConfig):
params.override({
'task': {
'train_data': {
'tf_data_service_address': flags_obj.tf_data_service,
},
'validation_data': {
'tf_data_service_address': flags_obj.tf_data_service,
}
}
})
# 4. Get the second level of override from `--params_override`.
# `--params_override` is typically used as a further override over the
......
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