… Medical research agent skills require safeguards beyond general-purpose evaluation, including scientific integrity, methodological validity, reproducibility, and boundary safety. …
… Safety concerns have rightly drawn growing research attention, and beneath them lie the values silently steering agent behavior. Existing value benchmarks, however, remain confined to LLMs, leaving agent values largely uncharted. …
… View arXiv page View PDF Add to collection Community For questions or model-evaluation requests, contact guijin.son@snu.ac.kr . objective evaluation을 하기에는 class imbalance가 너무 큰게 아닌게 싶네요. the way soohak treats refusal as a first-class signal is a clever move, highlighting a real frontier beyond pure… …
… However, existing UAV SAR research is dominated by traditional vision and path-planning methods and lacks a comprehensive and unified benchmark for embodied agents . …
… Key highlights from our research: 🏭 Grounded in Reality: 2,049 items based on Chinese national standards GB/T and real industrial product records evaluated across ZH, EN, RU, and VI . ⚠️ The "Overthinking" Trap: Surprisingly, we found that enabling extended reasoning thinking mode actually lowers s… …
… View arXiv page View PDF Project page GitHub 70 Add to collection Community 🎧👀 Audio-Visual Intelligence in Large Foundation Models: A Comprehensive Survey 📄 arXiv: 2605.04045 We are excited to release what we believe is the first comprehensive survey on Audio-Visual Intelligence AVI in the era of … …
… We introduce AcademiClaw, a bilingual benchmark of 80 complex, long-horizon tasks sourced directly from university students' real academic workflows -- homework, research projects, competitions, and personal projects -- that they found current AI agents unable to solve effectively. …
… The following papers were recommended by the Semantic Scholar API PIArena: A Platform for Prompt Injection Evaluation 2026 Claudini: Autoresearch Discovers State-of-the-Art Adversarial Attack Algorithms for LLMs 2026 Transient Turn Injection: Exposing Stateless Multi-Turn Vulnerabilities in Large L… …
… I found the explanation of Hugging Face’s “Skills Training” initiative — how it lets you use a coding‑agent like Claude Code or other supported agents to fine‑tune large language models, submit GPU jobs, monitor progress and push trained models to the Hub — particularly eye‑opening. …
… I guess another perk would be the safety alignment. …