TensorBoard logging requires TensorBoard with Python summary writer installed. This should be available in 1.14 or above

[LIMr1209]

Traceback (most recent call last):
File "/home/tian/.conda/envs/jiqi/lib/python3.6/site-packages/torch/utils/tensorboard/init.py", line 2, in
from tensorboard.summary.writer.record_writer import RecordWriter # noqa F401
File "/home/tian/Desktop/ai/machine-learn/Image_recognition/tensorboard.py", line 3, in
from torch.utils.tensorboard import SummaryWriter
ImportError: cannot import name 'SummaryWriter'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "/home/tian/Desktop/ai/machine-learn/Image_recognition/tensorboard.py", line 3, in
from torch.utils.tensorboard import SummaryWriter
File "/home/tian/.conda/envs/jiqi/lib/python3.6/site-packages/torch/utils/tensorboard/init.py", line 4, in
raise ImportError('TensorBoard logging requires TensorBoard with Python summary writer installed. '
ImportError: TensorBoard logging requires TensorBoard with Python summary writer installed. This should be available in 1.14 or above

[LIMr1209]

tensorboard --logdir=runs
TensorBoard 1.14.0a20190504 at http://tian-ORC:6006/ (Press CTRL+C to quit)

[gaoyang07]

tensorboard --logdir=runs
TensorBoard 1.14.0a20190504 at http://tian-ORC:6006/ (Press CTRL+C to quit)

It seems that you worked? If not, maybe this link https://www.endtoend.ai/pytorch-tensorboard may help you.

[LIMr1209]

@GeniusGaryant Thank you very much 😊

[ssnl]

cc @orionr The error message is confusing. Python summary writer isn't in TensorBoard 1.14.0a20190301 (it is in TensorBoard 1.14.0a20190505)

[orionr]

1.14 hasn't been released yet (the above are just master commits), but will be in the near future. This is what we reference in the message. For now, people should install with pip install tb-nightly until the 1.14 release.

[Maghoumi]

In case you've already tried pip install tb-nightly but still get the error message about TensorBoard 1.14, it could be that the TensorBoard packages in your ~/.local are older than 1.14.

This can happen if you install packages via pip install --user ... and some other packages via sudo pip install ..., which is known to mess things up.

[jphdotam]

Also, you may have the latest tb-nightly but you will wrongly get this error message if you're missing google's protobuf.

conda install -c anaconda protobuf
conda install future

Fixed it for me.

[visionscaper]

Next to pip install tb-nightly, I had to pip install future.

[naqute]

I think it's a version problem.
just run this:
pip install tensorboard==1.14.0
(not pip install tensorboard==1.14)

[goudantongxue]

Also, you may have the latest tb-nightly but you will wrongly get this error message if you're missing google's protobuf.

conda install -c anaconda protobuf
conda install future

Fixed it for me.

How to install tb-nightly? i found conda don't have this command. Do I have to use pip install tb-nightly?

[syomantak]

I think it's a version problem.
just run this:
pip install tensorboard==1.14.0
(not pip install tensorboard==1.14)

@naqute Does this work for gpu systems which are incompatible with tensorflow 1.14? My CUDA version won't support tf 1.14 :(

[naqute]

Sorry, my cuda is compatible with tf 1.14.0. I can not answer your question.

…

---Original--- From: "Syomantak Chaudhuri"<[email protected]&gt; Date: Sun, Apr 26, 2020 19:53 PM To: "pytorch/pytorch"<[email protected]&gt;; Cc: "Mention"<[email protected]&gt;;"NaQ"<[email protected]&gt;; Subject: Re: [pytorch/pytorch] TensorBoard logging requires TensorBoard with Python summary writer installed. This should be available in 1.14 or above (#20140) I think it's a version problem. just run this: pip install tensorboard==1.14.0 (not pip install tensorboard==1.14) @naqute Does this work for gpu systems which are incompatible with tensorflow 1.14? My CUDA version won't support tf 1.14 :( — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe.

[vtasStu]

tensorboard --logdir=runs
TensorBoard 1.14.0a20190504 at http://tian-ORC:6006/ (Press CTRL+C to quit)

It seems that you worked? If not, maybe this link https://www.endtoend.ai/pytorch-tensorboard may help you.

THX