Skip to main content

Python on Windows 10

Last weekend I wanted to clean up my python installation on Windows 10. In the past, I used a combination of Anaconda, Python app from the Windows Store, and Python distributions from python.org. Most of these were older versions that I have not updated in a while since I spend most of my time in WSL.

I started by deleting everything and cleaning up the installs. Afterwards, I downloaded python 3.8 from python.org, and proceeded with the installation. I selected install python for everyone and adding it to the %PATH% environment variable.

After the installation completed everything worked like a charm, until I started installing packages.

I started with installing TensorFlow:

pip install tensorflow

and got a cryptic error that one of the TensorFlow files was missing.

A quick internet search revealed that it is due to a missing setting for long filenames in Windows 10.  Following the instructions in this Stackoverflow post to enable long filenames option in the local computer policy, resolved the issue, and the installation proceeded smoothly.

Then I installed keras, and jupyterlab

pip install keras jupyterlab

and tried to run jupyter lab, which failed because Jupyter was not in the %PATH%.  Instead of adding another directory to the %PATH% variable, I ran jupyterlab through:

python -m jupyterlab

 

More keystrokes but works uniformly for any python module.

Now with python and the ML frameworks installed on Windows 10, I can make use of the GPU again. But not for long, since WSL is planning on adding GPU support soon, and there will be no need to live outside of WSL anymore.

Labels:

Comments

Post a Comment

Popular posts from this blog

Emacs on WSL2: From Monochrome Misery to Modern Elegance

Windows Subsystem for Linux (WSL) has come a long way—especially under Windows 11 . WSL2 now offers smooth integration for Linux graphical applications, making it feel less like a compatibility layer and more like a native experience. But if you're an Emacs user, you might have noticed something off. Launching Emacs under WSL can feel like stepping into a time machine. Tiny fonts, washed-out visuals, and a UI that evokes the green-and-amber glow of vintage terminals. Functional? Yes. Pleasant? Absolutely not. But here's the good news: it is easy to make Emacs under WSL2 look just as sharp and modern as it does on Mac OSX . The emacs-pgtk build is designed for better graphical integration under WSL. It uses the Pure GTK interface , which plays nicely with WSL’s GUI support. sudo apt install emacs-pgtk To make Emacs look great, we’ll use Windows’ rich font library. First, edit your font configuration: sudo emacs /etc/fonts/fonts.conf and add the Windows Font directory...
Post a Comment
Read more

MacOS Catalina, OneDrive, and case sensitive file systems

Over the weekend, I dusted off my old Macbook Air to search for some old family photos. I have not used the laptop for a long time, and it was completely out of charge. I plugged it in, and it quickly booted. Shortly after, I got bombarded with notifications that many of the applications needed updating, and that a new version of the OS was available.   I waited till I found the photos I was looking for, before attempting to upgrade anything. I also wanted to install OneDrive to get my old files to the cloud, so that I can access them from any of my devices, instead of dusting off old computers to get to them. The MacOS upgrade experience has always been fantastic, and this one was no different. The OS upgrade files downloaded quickly and after a restart and a quick install, the Macbook Air was ready to go.   Upgrading the installed applications was also a breeze, however in the process I discovered that a large majority of the applications installed were not compatible ...
Post a Comment
Read more

Randomized Algorithms: Polynomial equivalence

We are all accustomed to deterministic algorithms; we work with them every day, and feel comfortable in knowing that the results of running them are predictable, barring a coding error of course. The idea of randomized algorithms feels remote and uncomfortable, despite their usefulness and elegance.  There are a couple of great examples in the introductory chapters of the book " Probability and Computing " that are an eye opener. One is verifying polynomial identities: how can you tell that two different representations of polynomials are the same? For example, if we have two polynomials $P(x)$ and $Q(x)$, both of degree $d$ described by the following formulas: \[ P(x) = \Sigma_{i=0}^{i=d} a_i x^i \\ Q(x) = \Pi_{i=1}^{i=d} (x-b_i) \] how can we determine that they are the same polynomial? Intuitively we first check that the degrees are the same, then we could try to transform one form into the other, either by multiplying out the terms for $Q(x)$, collecting like t...
Post a Comment
Read more