rasbt / LLMs-from-scratch

Implement a ChatGPT-like LLM in PyTorch from scratch, step by step
https://www.amazon.com/Build-Large-Language-Model-Scratch/dp/1633437167
Other
32.84k stars 3.95k forks source link

Local setup: Suggestions to improve section about conda #348

Closed navneethc closed 2 months ago

navneethc commented 2 months ago

Bug description

Source of issue: https://github.com/rasbt/LLMs-from-scratch/tree/main/setup/01_optional-python-setup-preferences#1-download-and-install-miniforge

At the end of first step, it is mentioned:

If you work with many packages, Conda can be slow because of its thorough but complex dependency resolution process and the handling of large package indexes and metadata. To speed up Conda, you can use the following setting, which switches to a more efficient Rust reimplementation for solving dependencies:

conda config --set solver libmamba

A few things to a note:

What operating system are you using?

None

Where do you run your code?

None

Environment

rasbt commented 2 months ago

Thanks for the comment! Ah yes, setting libmamba manually shouldn't be necessary anymore then, and I can take out this section because it's redundant now. (Not sure why I thought of Rust...). Anyways, thanks for letting me know, I updated it.