未验证 提交 82e783e3 编写于 作者: T Taylor Robie 提交者: GitHub

bring NCF to l2 logging compliance (#5642)

上级 8a15a4df
...@@ -507,6 +507,7 @@ def instantiate_pipeline(dataset, data_dir, batch_size, eval_batch_size, ...@@ -507,6 +507,7 @@ def instantiate_pipeline(dataset, data_dir, batch_size, eval_batch_size,
# will then enter a loop waiting for the flagfile to be written. Once we see # will then enter a loop waiting for the flagfile to be written. Once we see
# that the async process has signaled that it is alive, we clear the system # that the async process has signaled that it is alive, we clear the system
# caches and begin the run. # caches and begin the run.
mlperf_helper.ncf_print(key=mlperf_helper.TAGS.RUN_CLEAR_CACHES)
mlperf_helper.clear_system_caches() mlperf_helper.clear_system_caches()
mlperf_helper.ncf_print(key=mlperf_helper.TAGS.RUN_START) mlperf_helper.ncf_print(key=mlperf_helper.TAGS.RUN_START)
write_flagfile(flags_, ncf_dataset) write_flagfile(flags_, ncf_dataset)
......
google-api-python-client>=1.6.7 google-api-python-client>=1.6.7
google-cloud-bigquery>=0.31.0 google-cloud-bigquery>=0.31.0
kaggle>=1.3.9 kaggle>=1.3.9
mlperf_compliance==0.0.8 mlperf_compliance==0.0.10
numpy numpy
oauth2client>=4.1.2 oauth2client>=4.1.2
pandas pandas
......
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