提交 7e83c5a5 编写于 作者: M mindspore-ci-bot 提交者: Gitee

!36 Fix user defined related comments

Merge pull request !36 from kouzhenzhong/comments_fix
...@@ -41,9 +41,9 @@ def get_summary_lineage(summary_dir, keys=None): ...@@ -41,9 +41,9 @@ def get_summary_lineage(summary_dir, keys=None):
summary_dir (str): The summary directory. It contains summary logs for summary_dir (str): The summary directory. It contains summary logs for
one training. one training.
keys (list[str]): The filter keys of lineage information. The acceptable keys (list[str]): The filter keys of lineage information. The acceptable
keys are `metric`, `hyper_parameters`, `algorithm`, `train_dataset`, keys are `metric`, `user_defined`, `hyper_parameters`, `algorithm`,
`model`, `valid_dataset` and `dataset_graph`. If it is `None`, all `train_dataset`, `model`, `valid_dataset` and `dataset_graph`.
information will be returned. Default: None. If it is `None`, all information will be returned. Default: None.
Returns: Returns:
dict, the lineage information for one training. dict, the lineage information for one training.
...@@ -112,10 +112,12 @@ def filter_summary_lineage(summary_base_dir, search_condition=None): ...@@ -112,10 +112,12 @@ def filter_summary_lineage(summary_base_dir, search_condition=None):
directories generated by training. directories generated by training.
search_condition (dict): The search condition. When filtering and search_condition (dict): The search condition. When filtering and
sorting, in addition to the following supported fields, fields sorting, in addition to the following supported fields, fields
prefixed with `metric/` are also supported. The fields prefixed with prefixed with `metric/` and `user_defined/` are also supported.
`metric/` are related to the `metrics` parameter in the training For example, the field should be `metric/accuracy` if the key
script. For example, if the key of `metrics` parameter is of `metrics` parameter is `accuracy`. The fields prefixed with
`accuracy`, the field should be `metric/accuracy`. Default: None. `metric/` and `user_defined/` are related to the `metrics`
parameter in the training script and user defined information in
TrainLineage/EvalLineage callback, respectively. Default: None.
- summary_dir (dict): The filter condition of summary directory. - summary_dir (dict): The filter condition of summary directory.
......
...@@ -56,6 +56,8 @@ class TrainLineage(Callback): ...@@ -56,6 +56,8 @@ class TrainLineage(Callback):
raise_exception (bool): Whether to raise exception when error occurs in raise_exception (bool): Whether to raise exception when error occurs in
TrainLineage. If True, raise exception. If False, catch exception TrainLineage. If True, raise exception. If False, catch exception
and continue. Default: False. and continue. Default: False.
user_defined_info (dict): User defined information. Only flatten dict with
str key and int/float/str value is supported. Default: None.
Raises: Raises:
MindInsightException: If validating parameter fails. MindInsightException: If validating parameter fails.
...@@ -233,6 +235,8 @@ class EvalLineage(Callback): ...@@ -233,6 +235,8 @@ class EvalLineage(Callback):
raise_exception (bool): Whether to raise exception when error occurs in raise_exception (bool): Whether to raise exception when error occurs in
EvalLineage. If True, raise exception. If False, catch exception EvalLineage. If True, raise exception. If False, catch exception
and continue. Default: False. and continue. Default: False.
user_defined_info (dict): User defined information. Only flatten dict with
str key and int/float/str value is supported. Default: None.
Raises: Raises:
MindInsightException: If validating parameter fails. MindInsightException: If validating parameter fails.
......
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