Different labels, same pitfalls. This position paper argues that reward models (for RL-based LLM training) and evaluation metrics face overlapping challenges—spurious correlations, reward hacking, data quality, and meta-evaluation. In some tasks, metrics even outperform reward models. Why it matters: Aligning these research communities could improve preference elicitation, robustness to
LLM agents work better together—with the right prompts. This work enhances the CoELA framework for collaborative embodied agents, testing multiple LLMs and prompt strategies to boost cooperation and decision-making in shared virtual spaces. The best combo improved efficiency with Gemma3 by 22% over the original setup. They also added
When chains get long, LLMs lose the thread. Self-Anchor pins it down. Self-Anchor structures a plan from the reasoning trajectory and automatically aligns the model’s attention to the most relevant steps while generating. Across six benchmarks, it outperforms state-of-the-art prompting and narrows the gap between general and specialized reasoning
Counterintuitive, but it works: add tiny shifts to boost robustness. This training-free, architecture-agnostic method uses stochastic resonance: apply small image translations, align features, aggregate, and map back—no extra modules, no attack-specific tuning. It recovers up to 68.1% of accuracy loss in classification, 71.9% in stereo, and 29.
Fine-grained spine reasoning needs fine-grained data. SpineMed introduces SpineMed-450k—the first large-scale dataset explicitly designed for vertebral-level reasoning across X-ray, CT, and MRI—plus SpineBench, a clinically grounded benchmark. Using clinician-in-the-loop curation and traceable instructions, the ecosystem supports Q&A, multi-turn consultations, and report generation. Why it matters: Evaluations
Some say synthetic CSAM has no victims. This paper shows why that’s dangerously wrong. The authors examine how AI-generated child sexual abuse material can revictimize known survivors, create synthetic depictions of children who were not abused, facilitate grooming and extortion, normalize exploitation, and lower barriers to offending. They caution
Fakes are getting scary good. This system fights back—smartly. UniShield is a multi-agent framework that detects and pinpoints image forgeries across manipulation, documents, DeepFakes, and AI-generated pics. Think of it like an airport with a smart dispatcher (perception agent) directing each bag to the right scanner (expert detectors), then
Plain-English explainer: Here’s what this new research means without technical jargon. In simple terms, the authors explore a problem that shows up in real systems and propose a way to handle it. Think of it like tuning a radio: with the right adjustment, the signal becomes clearer and small
Plain-English explainer: Here’s what this new research means without technical jargon. In simple terms, the authors explore a problem that shows up in real systems and propose a way to handle it. Think of it like tuning a radio: with the right adjustment, the signal becomes clearer and small
Plain-English explainer: Here’s what this new research means without technical jargon. In simple terms, the authors explore a problem that shows up in real systems and propose a way to handle it. Think of it like tuning a radio: with the right adjustment, the signal becomes clearer and small
Plain-English explainer: Here’s what this new research means without technical jargon. In simple terms, the authors explore a problem that shows up in real systems and propose a way to handle it. Think of it like tuning a radio: with the right adjustment, the signal becomes clearer and small
Plain-English explainer: Here’s what this new research means without technical jargon. In simple terms, the authors explore a problem that shows up in real systems and propose a way to handle it. Think of it like tuning a radio: with the right adjustment, the signal becomes clearer and small