Show HN: Trainlint, static checks for ML bugs that don't crash

Category: devtools

Tags: ml, static-analysis, linter, pytorch, data-leakage

Score: 6.0/10 (Innovation: 6, Technical: 5, Documentation: 8, Utility: 5)

trainlint is a static checker (AST-based) for common silent ML bugs like data leakage, missing eval mode, and gradient accumulation errors, paired with agent skills. It's interesting because it targets training-logic correctness that general linters miss, while staying dependency-free and offline.

Target audience: ML engineers, data scientists, backend devs writing PyTorch training code

Repository: https://github.com/Satvik-jain/trainlint · Python · MIT · 1 stars

View on Hacker News