TensorFlow：如何启动 Profiler

Posted 2023-03-06

【中文标题】TensorFlow：如何启动 Profiler

【英文标题】：TensorFlow: how to start the Profiler

【发布时间】：2018-11-26 12:30:01

【问题描述】：

我想分析我的模型。这是一个关于如何做到这一点的教程：https://towardsdatascience.com/howto-profile-tensorflow-1a49fb18073d。但我想使用 TensorFlow 分析器，如 https://github.com/tensorflow/tensorflow/blob/master/tensorflow/core/profiler/README.md#quick-start 所示。根据这篇文章，以下代码应该启动分析器：

# When using high-level API, session is usually hidden.
#
# Under the default ProfileContext, run a few hundred steps.
# The ProfileContext will sample some steps and dump the profiles
# to files. Users can then use command line tool or Web UI for
# interactive profiling.
with tf.contrib.tfprof.ProfileContext('/tmp/train_dir') as pctx:
  # High level API, such as slim, Estimator, etc.
  train_loop()

bazel-bin/tensorflow/core/profiler/profiler \
--profile_path=/tmp/train_dir/profile_xx
tfprof> op -select micros,bytes,occurrence -order_by micros

# To be open sourced...
bazel-bin/tensorflow/python/profiler/profiler_ui \
--profile_path=/tmp/profiles/profile_1

我生成了文件 profile_100 并找到了目录分析器。这就是我在终端中输入的内容：

bazel-/Users/mencia/anaconda/envs/tensorflow_py36/lib/python3.6/site-packages/tensorflow/profiler \ 
--profile_path=~/tmp/train_dir/profile_100

这引发了以下错误：

-bash: 
bazel-/Users/mencia/anaconda/envs/tensorflow_py36/lib/python3.6/site- packages/tensorflow/profiler: No such file or directory

我的目录分析器包含：

__init__.py
__pycache__

但是根据上面的代码，应该有

profiler/profiler

我没有。

如何启动 Profiler？

【参考方案1】：

您必须先构建分析器。克隆 TensorFlow 存储库 (git clone https://github.com/tensorflow/tensorflow.git) 并在根目录中运行 bazel build --config opt tensorflow/core/profiler:profiler。