Skip to content

Create a module that exposes all first-party plugin summaries #561

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Closed
wchargin opened this issue Sep 23, 2017 · 0 comments
Closed

Create a module that exposes all first-party plugin summaries #561

wchargin opened this issue Sep 23, 2017 · 0 comments
Assignees
@wchargin
Labels
core:backend

Comments

@wchargin
Copy link
Contributor

A user who wishes to use the new-style summaries will currently have to write,

from tensorboard.plugins.histograms import summary as histogram_summary

foo = histogram_summary.op(…)

(The import alias is necessary in the common case where the user wants to use summaries of multiple types.)

This is a significant increase in verbosity over the status quo, which only requires

import tensorflow as tf

foo = tf.summary.histogram(…)

Per previous discussion, we'd like to keep the main tensorboard module a normal module (no __init__ file) for readability and tooling purposes, but include a module tensorboard.summary that imports the summary op and pb files from individual plugins. The API would be

from tensorboard import summary

foo = summary.histogram(…)
foo_pb = summary.histogram_pb(…)
@wchargin wchargin self-assigned this Sep 23, 2017
wchargin added a commit that referenced this issue Sep 23, 2017
@wchargin
Add the tensorboard.summary public entry point
f1f0712 
Summary:
Resolves #561.

Test Plan:
Ran the `build_pip_package.sh` script to build the Python 2 package,
then created a new Python 2 virtualenv in which I installed only the
latest nightly TensorFlow release (2017-09-22) and the generated wheel.
Then wrote
```py
import tensorflow as tf
from tensorboard import summary
result = tf.Session().run(summary.scalar('shelby', tf.constant(3)))
s = tf.Summary()
s.ParseFromString(result)
print(s)
```
and verified that the result was a proper scalar summary.

I then repeated the process using `pip install tensorflow==1.3.0`
instead of the nightly version to ensure that the module works for
normal TensorFlow users. (Note: the audio summary will not work because
TensorFlow 1.3 lacks 22730fd4c633a74e59c03ff76dc92e6ae2d5d020. This is
fine; users can continue to use the old API for that one.)

wchargin-branch: summary-module
@wchargin wchargin mentioned this issue Sep 23, 2017
Merged
wchargin added a commit that referenced this issue Sep 26, 2017
@wchargin
Add the tensorboard.summary public entry point
4db1c24 
Summary:
Resolves #561.

Test Plan:
Ran the `build_pip_package.sh` script to build the Python 2 package,
then created a new Python 2 virtualenv in which I installed only the
latest nightly TensorFlow release (2017-09-22) and the generated wheel.
Then wrote
```py
import tensorflow as tf
from tensorboard import summary
result = tf.Session().run(summary.scalar('shelby', tf.constant(3)))
s = tf.Summary()
s.ParseFromString(result)
print(s)
```
and verified that the result was a proper scalar summary.

I then repeated the process using `pip install tensorflow==1.3.0`
instead of the nightly version to ensure that the module works for
normal TensorFlow users. (Note: the audio summary will not work because
TensorFlow 1.3 lacks 22730fd4c633a74e59c03ff76dc92e6ae2d5d020. This is
fine; users can continue to use the old API for that one.)

wchargin-branch: summary-module
wchargin added a commit that referenced this issue Sep 26, 2017
@wchargin
Add the tensorboard.summary public entry point (#562)
ab6ffbb 
Summary:
Resolves #561.

Test Plan:
Ran the `build_pip_package.sh` script to build the Python 2 package,
then created a new Python 2 virtualenv in which I installed only the
latest nightly TensorFlow release (2017-09-22) and the generated wheel.
Then wrote
```py
import tensorflow as tf
from tensorboard import summary
result = tf.Session().run(summary.scalar('shelby', tf.constant(3)))
s = tf.Summary()
s.ParseFromString(result)
print(s)
```
and verified that the result was a proper scalar summary.

I then repeated the process using `pip install tensorflow==1.3.0`
instead of the nightly version to ensure that the module works for
normal TensorFlow users. (Note: the audio summary will not work because
TensorFlow 1.3 lacks 22730fd4c633a74e59c03ff76dc92e6ae2d5d020. This is
fine; users can continue to use the old API for that one.)

wchargin-branch: summary-module
jart pushed a commit to jart/tensorboard that referenced this issue Sep 26, 2017
@wchargin @jart
Add the tensorboard.summary public entry point (tensorflow#562)
c0a0e19 
Summary:
Resolves tensorflow#561.

Test Plan:
Ran the `build_pip_package.sh` script to build the Python 2 package,
then created a new Python 2 virtualenv in which I installed only the
latest nightly TensorFlow release (2017-09-22) and the generated wheel.
Then wrote
```py
import tensorflow as tf
from tensorboard import summary
result = tf.Session().run(summary.scalar('shelby', tf.constant(3)))
s = tf.Summary()
s.ParseFromString(result)
print(s)
```
and verified that the result was a proper scalar summary.

I then repeated the process using `pip install tensorflow==1.3.0`
instead of the nightly version to ensure that the module works for
normal TensorFlow users. (Note: the audio summary will not work because
TensorFlow 1.3 lacks 22730fd4c633a74e59c03ff76dc92e6ae2d5d020. This is
fine; users can continue to use the old API for that one.)

wchargin-branch: summary-module
jart pushed a commit that referenced this issue Sep 26, 2017
@wchargin @jart
Add the tensorboard.summary public entry point (#562)
ae61422 
Summary:
Resolves #561.

Test Plan:
Ran the `build_pip_package.sh` script to build the Python 2 package,
then created a new Python 2 virtualenv in which I installed only the
latest nightly TensorFlow release (2017-09-22) and the generated wheel.
Then wrote
```py
import tensorflow as tf
from tensorboard import summary
result = tf.Session().run(summary.scalar('shelby', tf.constant(3)))
s = tf.Summary()
s.ParseFromString(result)
print(s)
```
and verified that the result was a proper scalar summary.

I then repeated the process using `pip install tensorflow==1.3.0`
instead of the nightly version to ensure that the module works for
normal TensorFlow users. (Note: the audio summary will not work because
TensorFlow 1.3 lacks 22730fd4c633a74e59c03ff76dc92e6ae2d5d020. This is
fine; users can continue to use the old API for that one.)

wchargin-branch: summary-module
@wchargin wchargin mentioned this issue Sep 27, 2017
Closed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees

@wchargin wchargin

Labels
core:backend
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
1 participant
@wchargin