scicode-lint

[cwhanse]

This is interesting: https://github.com/authentic-research-partners/scicode-lint

I wonder what it would find in the pvlib-python codebase, and what it would take to provide a PV modeling methodology.

I got lost in the setup/configuration of the required LLM on my Windows machine - anyone have a Ubuntu setup that would like to try this out?

[echedey-ls]

Looks interesting, thanks for sharing.

Built for scientists applying ML - biology, chemistry, physics, neuroscience, engineering, and beyond. You're an expert in your domain; catching ML methodology bugs isn't your day job. This tool fills that gap.
[https://github.com/authentic-research-partners/scicode-lint#the-problem]

Seems specifically appropriate for codebases where you train/test/benchmark ML models. Surely useful for many PV projects, but pvlib is not in this category (for the moment?). May it be able to check for other common non-ML mistakes, thou?

GPU with 16GB+ VRAM and native FP8 support (RTX 4060 Ti 16GB, RTX 4070+, RTX 4090, L4, etc.)
https://github.com/authentic-research-partners/scicode-lint#prerequisites

I have 2GB lmao.