PROBE benchmark exposes gaps in AI protein-function annotation
A 1,000-protein, five-organism benchmark found that leading open-source language models showed biological semantic alignment but achieved a best macro-F1 of ...
A 1,000-protein, five-organism benchmark found that leading open-source language models showed biological semantic alignment but achieved a best macro-F1 of ...
© 2025 Scienmag - Science Magazine
© 2025 Scienmag - Science Magazine