RAGEN ranks 3rd hardest in FrontierCS 2: lift a 0.5B model with 0.5 A100 hour
wzenus · x · 2026-10-30
RAGEN is ranked the 3rd hardest task in FrontierCS 2, built from code the author hand-wrote in January 2025, before coding agents took off. He notes the bar is admittedly high: improving a 0.5B model's performance via RL with just 0.5 A100 GPU hours, coding time included.
More from Research
- Investor's Last Exam Benchmark: Top AI Models Ace Research but Fail at Investment Judgment — Huangfei Jiang · 2026-10-30
- SpecSlot Boosts Pixel-Space Image Generation: +2.84 FID on ImageNet-512 at Lower Cost — Jiongfan Yang · 2026-10-30
- Apple Trains UI Navigation Agents with Online RL Directly on Live iOS Apps — apple · 2026-10-30
- KAIST's L3G-World Generates Infinite Interactive Worlds in Real Time with Latent 3D Gaussian Memory — kaist-ai · 2026-10-30
- PhysInsert from Beihang and Meitu Achieves Physically Plausible Reference-Based Object Insertion — Yue Gong · 2026-10-30
- AB-Attention Speeds Up Video Diffusion Transformers with Training-Free Accurate Block-Sparse Attention — Bin Xu · 2026-10-30