Installing the TensorFlow Profiler UI

TensorFlow Profiler UI

1. Install Python Dependencies.

   pip install --user -r requirements.txt

2. Install pprof.

Note: Installation requires the first installation of go and Google home pprof tools.

3. Create a profile context.

4. Start the UI.

   python ui.py --profile_context_path=/your path/

5. Open URLs through the browser to view the UI interface

Error:

OSError: [Errno 2] No such file or directory

After installing the Pprof

 OSError: [Errno 13] Permission denied

Ubuntu System Installation Go

1. Since this machine uses Xshell to connect to a Linux system remotely, and this machine does not have a wall, you cannot use the following command:

   wget -c https://storage.googleapis.com/golang/go1.7.1.linux-amd64.tar.gz

But opening the above URL directly can download the TAR package natively, so download the tar package locally and then upload it to the remote.

2. Upload files to Ubuntu via Xshell

Installing Lrzsz:apt-get Install Lrzsz

Then drag-and-drop the downloaded tar on this xshell.

3. Set environment variables after decompression:

   tar -C /usr/local -xzf go1.7.1.linux-amd64.tar.gz
cd ~
vim .profile

Not added at the end of the. Profile file:

export PATH=$PATH:/usr/local/go/bin

Save exit to make the statement effective and test to see if the installation was successful:

. .profile

Installing the Pprof tool

go get -u github.com/google/pprof

The above command will be faulted

package github.com/google/pprof: cannot download, $GOPATH not set. For more details see: go help gopath

Here's how to fix it:

env GOPATH=/usr/local/go/bin/pprof go get -u github.com/google/pprof

Manually make Gopath before installation, this path is also the PPROF installation path!