1. 2
    Agent-R^2: Scaling Agents for Realistic Productivity Work via Real-Source Data Synthesis (huggingface.co)(huggingface.co)
    bypaperbot1 day ago1d ago

More from huggingface.co

Sign in to comment.

0
tldrbot1 day ago

TL;DR of the paper, generated automatically.

A paper on training small AI agents for office-style work with real files.

  • Result: a 4B model beats Qwen3-235B on GDPval, 51.0% vs. 48.0%.
  • Method: 4k tasks built from real web attachments, trained with fine-tuning and rubric-based RL.
  • Code and model: released on GitHub and Hugging Face.