pyenv VS mamba

For more information visit: pyenv repository

Note that virtual environments assume you are using the same global version of Python. Often, this is not the case and additional tools like pyenv can be used alongside virtual environments when you need to switch between versions of Python itself on your local machine.

Python version manager pyenv

First of all, install the Python virtual environment from these links: 1 and 2. I developed my GPT-based API in Python version 3.8.18. Pick any Python versions >= 3.7.

Note: Most modern systems will probably have a system environment that meets this requirement, but if yours does not or if you prefer not to install anything in your system environment (even if it's just PDM) check out asdf or pyenv to help install and manage additional Python environments.

When dealing with software development, reproducibility is key. This is why we encourage you to use Python virtual environments to set up an isolated environment for your project. Virtual environments allow the isolation of dependencies, which plays a crucial role to avoid breaking compatibility between different projects. We cannot cover all the details about virtual environments in this post, but we encourage you to learn more about venv, pyenv or conda for a better understanding on how to create and manage virtual environments.

For Python, with your use case I would avoid system packages, no matter the distro. It sounds like it would be worth setting up pyenv and working exclusively with virtual environments.

For OSX there is homebrew or pyenv (pyenv is another solution on Linux). As pyenv compiles from source it will require setting up XCode (the Apple IDE) tools to support this which can be pretty bulky. Windows users have chocolatey but the issue there is it works off the binaries. That means it won't have the latest security release available since those are source only. Conda is also another solution which can be picked up by Visual Studio Code as available versions of Python making development easier. In the end it might be best to consider using WSL on Windows for installing a Linux version and using that instead.

> how the recall can grow unbounded with no tradeoff

this? https://github.com/state-spaces/mamba/issues/175

If you want to learn this stuff as a computer engineer, you can read the code here [0]. I find the math quite helpful.

[0]: https://github.com/state-spaces/mamba

This model was discussed recently: https://news.ycombinator.com/item?id=38522428 It's a new kind of ML model architecture that can be used instead of a transformer in LLMs.

See also the original repo from the paper: https://github.com/state-spaces/mamba

[2] - https://github.com/state-spaces/mamba

Out of curiosity, does anyone feel as though there's any benefit to linking to reddit when we can link to whatever the link is? I for one do not click the link and read discussion on reddit - if I wanted that sort of discussion, I would browse there, not HN.

The library is compatible with any auto-regressive model, not transformers. To prove our point we integrated Mamba, a new state-space model architecture, to the library. Try it out!

I ran the NanoGPT of Karparthy replacing Self-Attention with Mamba on his TinyShakespeare Dataset and within 5 minutes it started spitting out the following:

You might have come across the paper Mamba paper in the last days, which was the first attempt at scaling up state space models to 2.8B parameters to work on language data.