Loading activity…
@gaelvaroquaux
Coder & Research director @inria ►Data, Health, & Computer science ►Python coder, (co)founder of @scikit_learn, joblib, @probabl_ai ►Art: @artgael ►Physics PhD
AI
The author says language-model editors often made strong abstracts worse and pointed readers to a COLM 2026 poster.
AI
A reader of a FAIR and NYU paper says models below perhaps 64 million parameters need intensive tuning—and warns against equating easier tuning at larger tested scales with simpler training.
AI
Researcher shares report on AI bypassing human verification checks.
AI
Gael Varoquaux cites lawsuits and Anthropic projects as examples of AI data practices.
AI
In an X reply, the scikit-learn contributor recounts a conversation about web data access.
AI
Explains agent cheating in evaluations by appealing to beliefs and goals.
