Developer duel
🔥 A's original project quality is a flat 0, with the entire score stacked on ecosystem impact, contribution quality and account activity — a classic open-source workhorse who can't even deliver a single decent original project, just grinding free contributions for VLMEval to farm data, with original ability completely crippled. B scored high on original project quality thanks to a self-made 2253-star Jupyter Notebook, but gets completely crushed by A on contribution quality and ecosystem impact: 6
1) For A: Stop acting as a PR QC inspector grinding for others. Turn your experience contributing to VLMEval into your own original project — add a proper README, tests, and a formal release to build an actually usable multimodal tool, instead of being a code monkey with only contributions and zero original output. 2) For B: Ditch the resume-decorating fork projects. Iterate on that 2253-star Notebook of yours, add full features, tests, and a formal release so it doesn't stay a one-off gimmick.