A PhD student at the Max Planck Institute working on AI safety with research focused on risks from automated AI research.
How media typically covers Ben Rank
Ben Rank as author
PostTrainBench demonstrates that AI agents can autonomously post-train small language models within constrained resources (10 hours, 1 H100 GPU), with dataset curation and constraint awareness being more important than training duration for performance improvements.
“Author of "PostTrainBench" in PostTrainBench Official Site”