Tensorboard: No train metrics show up in scalar plots

[zzb3886]

Consider Stack Overflow for getting support using TensorBoard—they have
a larger community with better searchability:

https://stackoverflow.com/questions/tagged/tensorboard

Do not use this template for for setup, installation, or configuration
issues. Instead, use the “installation problem” issue template:

https://github.com/tensorflow/tensorboard/issues/new?template=installation_problem.md

To report a problem with TensorBoard itself, please fill out the
remainder of this template.

Environment information (required)

Please run diagnose_tensorboard.py (link below) in the same
environment from which you normally run TensorFlow/TensorBoard, and
paste the output here:

https://raw.githubusercontent.com/tensorflow/tensorboard/master/tensorboard/tools/diagnose_tensorboard.py

Diagnostics

Diagnostics output

``````
--- check: autoidentify
INFO: diagnose_tensorboard.py version 4725c70c7ed724e2d1b9ba5618d7c30b957ee8a4

--- check: general
INFO: sys.version_info: sys.version_info(major=3, minor=6, micro=8, releaselevel='final', serial=0)
INFO: os.name: posix
INFO: os.uname(): posix.uname_result(sysname='Linux', nodename='master1', release='4.15.0-66-generic', version='#75-Ubuntu SMP Tue Oct 1 05:24:09 UTC 2019', machine='x86_64')
INFO: sys.getwindowsversion(): N/A

--- check: package_management
INFO: has conda-meta: False
INFO: $VIRTUAL_ENV: None

--- check: installed_packages
INFO: installed: tensorboard==2.0.1
INFO: installed: tensorflow==2.0.0
INFO: installed: tensorflow-estimator==2.0.1

--- check: tensorboard_python_version
INFO: tensorboard.version.VERSION: '2.0.1'

--- check: tensorflow_python_version
/usr/lib/python3/dist-packages/requests/__init__.py:80: RequestsDependencyWarning: urllib3 (1.25.3) or chardet (3.0.4) doesn't match a supported version!
RequestsDependencyWarning)
INFO: tensorflow.__version__: '2.0.0'
INFO: tensorflow.__git_version__: 'v2.0.0-rc2-26-g64c3d38'

--- check: tensorboard_binary_path
INFO: which tensorboard: b'/home/bz/.local/bin/tensorboard\n'

--- check: readable_fqdn
INFO: socket.getfqdn(): 'master1.bz'

--- check: stat_tensorboardinfo
INFO: directory: /tmp/.tensorboard-info
INFO: os.stat(...): os.stat_result(st_mode=16895, st_ino=14562209, st_dev=2430, st_nlink=2, st_uid=1000, st_gid=1000, st_size=4096, st_atime=1572563590, st_mtime=1573247067, st_ctime=1573247
067)
INFO: mode: 0o40777

--- check: source_trees_without_genfiles
INFO: tensorboard_roots (1): ['/home/bz/.local/lib/python3.6/site-packages']; bad_roots (0): []

--- check: full_pip_freeze
INFO: pip freeze --all:
absl-py==0.8.1
asn1crypto==0.24.0
astor==0.8.0
attrs==17.4.0
Automat==0.6.0
bleach==2.1.2
cachetools==3.1.1
certifi==2018.1.18
chardet==3.0.4
click==6.7
colorama==0.3.7
command-not-found==0.3
configobj==5.0.6
constantly==15.1.0
cryptography==2.1.4
decorator==4.1.2
distro-info===0.18ubuntu0.18.04.1
entrypoints==0.2.3.post1
eventkit==0.8.5
gast==0.2.2
google-auth==1.7.0
google-auth-oauthlib==0.4.1
google-pasta==0.1.8
grpcio==1.25.0
h5py==2.10.0
html5lib==0.999999999
httplib2==0.9.2
hyperlink==17.3.1
ib-insync==0.9.53
idna==2.6
incremental==16.10.1
ipykernel==4.8.2
ipython==5.5.0
ipython-genutils==0.2.0
ipywidgets==6.0.0
Jinja2==2.10
joblib==0.14.0
jsonschema==2.6.0
jupyter-client==5.2.2
jupyter-console==5.2.0
jupyter-core==4.4.0
Keras==2.3.1
Keras-Applications==1.0.8
Keras-Preprocessing==1.1.0
keyring==10.6.0
keyrings.alt==3.0
language-selector==0.1
Markdown==3.1.1
MarkupSafe==1.0
mistune==0.8.3
nbconvert==5.3.1
nbformat==4.4.0
nest-asyncio==1.0.0
netifaces==0.10.4
notebook==5.2.2
numpy==1.17.3
oauthlib==3.1.0
opt-einsum==3.1.0
PAM==0.4.2
pandas==0.24.2
pandocfilters==1.4.2
pexpect==4.2.1
pickleshare==0.7.4
pip==19.3.1
prompt-toolkit==1.0.15
protobuf==3.10.0
pyasn1==0.4.2
pyasn1-modules==0.2.1
pycrypto==2.6.1
Pygments==2.2.0
pygobject==3.26.1
pyOpenSSL==17.5.0
pyserial==3.4
python-apt==1.6.4
python-dateutil==2.8.0
python-debian==0.1.32
pytz==2019.1
pyxdg==0.25
PyYAML==3.12
pyzmq==16.0.2
requests==2.18.4
requests-oauthlib==1.3.0
requests-unixsocket==0.1.5
rsa==4.0
scikit-learn==0.21.3
scipy==1.3.1
SecretStorage==2.3.1
selenium==3.141.0
service-identity==16.0.0
setuptools==41.6.0
simplegeneric==0.8.1
six==1.13.0
sklearn==0.0
ssh-import-id==5.7
systemd-python==234
tdameritrade==0.0.7
tensorboard==2.0.1
tensorflow==2.0.0
tensorflow-estimator==2.0.1
termcolor==1.1.0
terminado==0.7
testpath==0.3.1
tornado==4.5.3
tqdm==4.32.2
traitlets==4.3.2
Twisted==17.9.0
ufw==0.36
unattended-upgrades==0.1
urllib3==1.25.3
vboxapi==1.0
wcwidth==0.1.7
webencodings==0.5
Werkzeug==0.16.0
wheel==0.33.6
wrapt==1.11.2
zope.interface==4.3.2

``````

For browser-related issues, please additionally specify:

• Browser type and version (e.g., Chrome 64.0.3282.140):
• Screenshot, if it’s a visual issue:

Issue description

I just upgraded tensorflow to 2.0. In training, I noticed tensorboard now has two runs for each experiment, including train and validation. However, only validation has scalar value curves. Train metric plots are always empty.

I can reproduce this issue by using the script in tensorboard get started guide: https://www.tensorflow.org/tensorboard/get_started. The script prints out reasonable train and val metrics as it should, but I'm just not getting the right plots.

[zzb3886]

I tried this a few more times. It looks like tensorboard 2.0 has trouble updating the train metrics by itself. If I kill tensorboard and restart it, it will then show both train and validation metrics. If the training is still ongoing, the validation metrics will be updated where as the train metrics are stuck.

[rmothukuru]

@zzb3886,
I ran the script provided in the link, https://www.tensorflow.org/tensorboard/get_started and could observe the Graphs for both Training and Validation. Here is the Gist.

Can you please provide more details about your issue.

Regarding If the training is still ongoing, the validation metrics will be updated where as the train metrics are stuck. =>

Tensorflow Graphs get updated from the Event Files stored during Training. So, it is recommended to see and analyze the graphs after the Training is completed, rather than during Training. Please let me know your opinion about the same.

[dgrahn]

I'm encountering the same problem. The train scaler isn't updated until TensorBoard is restarted.

[zzb3886]

In the script, if the tensorboard is started before training is started, then the problem occurs.

[dgrahn]

If I restart tensorboard during the training, the metrics get updated once,
but the problem persists.

On Thu, Nov 14, 2019, 1:04 PM zzb3886 [e-mail] wrote:

In the script, if the tensorboard is started before training is started,
then the problem occurs.

—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
https://github.com/tensorflow/tensorboard/issues/2911?email_source=notifications&email_token=AADALVMT64H2YJEOAA6YNPDQTWHKRA5CNFSM4JK64WF2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEECXXWI#issuecomment-554007513,
or unsubscribe
https://github.com/notifications/unsubscribe-auth/AADALVJTSQHAKIGEXVC2E7TQTWHKRANCNFSM4JK64WFQ
.

[wchargin]

Probably a duplicate of #2084; can you please try the workaround listed
in that issue and see if it resolves the problem?
https://github.com/tensorflow/tensorboard/issues/2084#issuecomment-483395808

[psybuzz]

Thanks for the report. I can confirm this was working with tf-nightly-2.0-preview==2.0.0.dev20190306 and broken in tf-nightly-2.0-preview==2.0.0.dev20190307. Bisected to https://github.com/tensorflow/tensorflow/commit/c66b603990b9404dc1eb57de9d595aa0ffc8197f

So it seems Keras callbacks have been affected by this bug since March, sadly. I'm going to triage this to someone who knows more context.

Googlers, see cl/237090182

[zzb3886]

Adding profile_batch=0 to the keras callback resolves it.

[psybuzz]

Duplicate of #2084

[AI-P-K]

I have the following code and I can't manage to get tensorboard to show my anything else but epoch_accuracy and epoch_loss. Can anyone help me? i have followed the steps above and is still not working.

This is the command I run in terminal tensorboard --logdir='logs/'

import tensorflow as tf
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, Dropout, Activation, Flatten, Conv2D, MaxPooling2D
import pickle
import numpy as np
from tensorflow.keras.preprocessing.image import ImageDataGenerator
import time
from datetime import datetime
from packaging import version
import os

X = pickle.load(open("X.pickle","rb"))
y = pickle.load(open("y.pickle","rb"))

X=np.array(X/255.0)
y=np.array(y)

dense_layers = [0, 1, 2]
layer_sizes = [32, 64, 128]
conv_layers = [1, 2, 3]

for dense_layer in dense_layers:
for layer_size in layer_sizes:
for conv_layer in conv_layers:
NAME = "{}-conv-{}-nodes-{}-dense-{}".format(conv_layer, layer_size, dense_layer, int(time.time()))
tensorboard = tf.keras.callbacks.TensorBoard(log_dir ='/users/silviumarc/pycharmprojects/classifier/logs/{}'.format(NAME), update_freq='epoch', profile_batch=0, histogram_freq=1)
print(NAME)

model = Sequential()
model.add(Conv2D(layer_size, (4,4), input_shape = X.shape[1:]))
model.add(Activation("relu"))
model.add(MaxPooling2D(pool_size=(2,2)))

for l in range(conv_layer-1):
model.add(Conv2D(layer_size, (4,4)))
model.add(Activation("relu"))
model.add(MaxPooling2D(pool_size=(2,2)))

model.add(Flatten())
for l in range(dense_layer):
model.add(Dense(layer_size))
model.add(Activation("relu"))

model.add(Dense(64))
model.add(Activation('relu'))

model.add(Dense(1))
model.add(Activation("sigmoid"))

model.compile(loss="binary_crossentropy",
optimizer= 'adam',
metrics=['accuracy'])

model.fit(X, y, batch_size=7, epochs=2, validation_split=0.5, callbacks=[tensorboard])