Nigel
/
mindspore
forked from mindspore-Ecosystem/mindspore
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

!3539 Change at-most collected tensor summary from 50 to 20 when auto-calculated 

Merge pull request !3539 from LiHongzhang/f50_t20_r
This commit is contained in:
mindspore-ci-bot 2020-07-27 19:54:19 +08:00 committed by Gitee
parent 49e8727d37 d86668d216
commit 67600c1d8c
1 changed files with 2 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
mindspore/train/callback/_summary_collector.py
View File

@ -110,7 +110,7 @@ class SummaryCollector(Callback):
lineage page. In the custom data, the key type support str, and the value type support str/int/float. lineage page. In the custom data, the key type support str, and the value type support str/int/float.
Default: None, it means there is no custom data. Default: None, it means there is no custom data.
collect_tensor_freq (Optional[int]): Same as the `collect_freq`, but controls TensorSummary specifically. collect_tensor_freq (Optional[int]): Same as the `collect_freq`, but controls TensorSummary specifically.
Default: None, which means the frequency is auto-calculated just to collect at most 50 steps TensorSummary. Default: None, which means the frequency is auto-calculated just to collect at most 20 steps TensorSummary.
max_file_size (Optional[int]): The maximum size in bytes each file can be written to the disk. max_file_size (Optional[int]): The maximum size in bytes each file can be written to the disk.
Default: None, which means no limit. Default: None, which means no limit.
@ -295,7 +295,7 @@ class SummaryCollector(Callback):
self._collect_dataset_graph(cb_params) self._collect_dataset_graph(cb_params)
if self._collect_tensor_freq is None: if self._collect_tensor_freq is None:
default_tensor_summary_limit = 50 default_tensor_summary_limit = 20
total_step = cb_params.epoch_num * cb_params.batch_num total_step = cb_params.epoch_num * cb_params.batch_num
self._collect_tensor_freq = max(self._collect_freq, total_step // default_tensor_summary_limit) self._collect_tensor_freq = max(self._collect_freq, total_step // default_tensor_summary_limit)