Daily AI Papers Briefing (2026-08-19)

Podliアプリで聴く 🎧

お気に入りの番組をフォローして、オフラインでも車内でもCarPlayやAndroid Autoで再生。いつでも続きから聴けます。無料でお試しできます。

【本日の論文】
1. HarnessEval-W: Agentifying the Evaluation of Visual Worlds
https://huggingface.co/papers/2608.16859
2. StateM: Reaching 95.3% Raw Accuracy, or a \$15 Frontier Run, on Terminal-Bench 2.1 via Harness Scaling
https://huggingface.co/papers/2608.15089
3. VibeWorlding: Can Multimodal Agents Construct 3D Open Worlds End-to-End?
https://huggingface.co/papers/2608.15265
4. Large Discovery Models: Empirically-grounded Model-Based Open-Ended Search
https://huggingface.co/papers/2608.15669
5. Learn What's Left, Not What's Mastered: Saturation Aware Advantage Reweighting for Multi-Reward Policy Optimization
https://huggingface.co/papers/2608.16072