The paper introduces the Language Model Agent Machine (LAM), a framework that quantifies the computational resources consumed by language‑model agents’ harnesses—such as bounded context, memory, tools, and verification. It presents four main theoretical results: (1) communication costs mapped to pebbling lower bounds, (2) memory‑access separations revealing a linear gap between random and non‑speculative access, (3) recomputation bounds for bit‑reversal DAGs that show when storing intermediate state is beneficial, and (4) reliability bounds that yield optimal verification intervals. Experiments on GPT‑6 Astra validate the theory’s predictions for communication, reliability, and trade‑offs in practical tasks.
By Zelin Zhao (Georgia Institute of Technology), Xinyu Guo (Georgia Institute of Technology), Jingyuan Zhang (Georgia Institute of Technology), Yuxuan Zhang (Etude AI), Yongxin Chen (Georgia Institute of Technology)
The article reviews the challenges and opportunities of detecting social bots amid the rise of advanced AI chatbots. It highlights gaps in current detection methods, especially regarding AI-generated conversations, and identifies emerging trends such as synthetic data generation, multimodal cross‑platform detection, low‑resource language support, and federated learning approaches. The authors propose these directions as promising avenues for future research.
By Emilio Ferrara
BioMol-MQA is a new question‑answering dataset focused on polypharmacy that combines a multimodal knowledge graph—containing both text and molecular structure—with challenging questions designed to test large language models’ ability to retrieve and reason over this diverse information. The dataset highlights the limitations of current retrieval‑augmented generation systems, which typically handle only single‑modality text, by demonstrating that existing LLMs perform poorly unless provided with the necessary multimodal background data. This underscores the need for more robust RAG frameworks capable of integrating multiple data types for accurate responses.
By Saptarshi Sengupta, Shuhua Yang, Paul Kwong Yu, Fali Wang, Suhang Wang
The paper introduces Percept-V, a dataset of 6,000 program-generated images across 30 domains designed to test simple visual perception skills from the TVPS-4 framework. Experiments show that state‑of‑the‑art multimodal large language models perform poorly compared to humans, especially as image complexity increases, and that fine‑tuning yields only limited generalization to related datasets. The study highlights specific perception skills that remain challenging for current models.
By Samrajnee Ghosh, Ashish Goswami, Naman Agarwal, Hemanshu Garg, Chinmay Mittal, Mausam, Parag Singla
The paper introduces a PyTorch library that unifies 55 hyperspectral image models across six deep‑learning paradigms, providing a common registry, automatic tensor adaptation, and standardized constructors. It includes 24 benchmark scenes from various sensors, with tools for data preprocessing, patch extraction, and rigorous train‑test partitioning to avoid overlap. Experiments on 1,320 model‑scene combinations show that scene difficulty outweighs architecture choice, with no single paradigm dominating and small models performing competitively with much larger ones.
By Tanishq Rachamalla, Aryan Das, Srishti Kaushik, Swalpa Kumar Roy
EditHero is presented as the first benchmark for long-horizon, part-level 3D editing, featuring natural-language instructions and target images for both geometry and texture. The benchmark uses a deterministic assembly engine that produces the exact target after each edit, and every sequence is manually reviewed. It compares non-agentic top‑down methods with LLM/VLM agent bottom‑up approaches, finding that the latter follow instructions more closely and preserve unedited parts better, though each edit takes minutes.
By Ruihan Yu, Yu-Ju Tsai, Muyao Niu, Runyi Li, Lian Fu, Hanqing Liu, Zheng-Hui Huang, Yonghao Yu, Sho Kuno, Ming-Hsuan Yang, Kaipeng Zhang, Zhixiang Wang
MEA is a multi‑agent framework that automates the selection and configuration of explanation tools for machine learning models across tabular, text, and vision data. A Proposer agent chooses appropriate tools based on the question and modality, while an Actor agent is trained end‑to‑end for faithfulness, producing natural‑language explanations. Experiments show that MEA outperforms existing post‑hoc explainers and closed‑source baselines, achieving significant faithfulness gains on six datasets.
By Yuyang Cheng, Raghav Kaushik Ravi, Srivarshinee Sridhar, Sriparna Saha, Akash Ghosh, Chirag Agarwal
The paper introduces a Mixture-of-Agents (MoA) approach to quantify the per-token compute required by large language models. By having a panel of fifteen models of varying sizes attempt to reproduce each token, the authors define the smallest successful agent’s inference cost as the token’s sufficient compute, providing an upper bound on necessary computation. Experiments on benchmarks show that a 0.5B model can reproduce most tokens, and that the MoA-derived compute map can reduce latency in model routing and drafting tasks while improving or maintaining accuracy.
By Zhixu Du, Weijia Han, Hai Helen Li, Yiran Chen
LUMOS is a diagnostic framework that tracks how knowledge in large language models (LLMs) originates from training data and manifests in outputs, using the fully transparent OLMo 2 corpus. The study finds that while models encode rare facts with high separability (84%), they often fail to express them behaviorally (54%), and self‑reflection accuracy drops sharply on unseen content. These results show that incorporating the training‑data axis into evaluation turns speculative claims into verifiable evidence, suggesting it should become a standard part of LLM knowledge assessment.
By Seoyeon Ye, Gayoung Kim, Jiyoung Hong, Sookyung Kim, Hyunsoo Cho
The study investigates how AI agents influence each other when they disagree, measuring persuasion as the change in an agent’s decision after a single exchange. Across seven open‑weight models and three language tasks, it finds that persuasion is strong—receivers often abandon their initial judgment after seeing a peer’s answer and explanation. Surprisingly, neither certainty nor model size reliably predicts persuasion dynamics; small models can persuade and resist larger ones just as effectively, and the shift depends more on the listener’s susceptibility than the speaker’s persuasiveness.
By Frida N{\o}hr Laustsen, Marie Haahr Petersen, Victoria Popa, Ariel Flint, Romualdo Pastor-Satorras, Andrea Baronchelli, Luca Maria Aiello
MOF-VERIFY is a failure-aware agentic harness designed to improve hypothesis verification for metal‑organic frameworks (MOFs). It introduces a diagnostic benchmark with four task families—structural grounding, synthesis‑condition verification, evidence‑sufficiency verification, and MLIP‑based computational verification—to pinpoint failures in knowledge access, evidence acquisition, and reasoning. Guided by these diagnostics, MOF‑Verify addresses structural, literature, evidence‑sufficiency, and computational bottlenecks, achieving significant performance gains over direct inference and retrieval‑based baselines across multiple large language models.
By Donghyun Lee, Taehoon Lee, Geonhee Ahn, Jieun Kim, Jihyun Park, Suyeon Cho, Yoona Kim, Chaerim Shin, Hoi Ri Moon, Jonggeol Na, Sukho Hong, Jihwan Oh, Soo Kyung Kim
The paper presents an expert‑validated benchmark of 1,042 items across ten areas of Colombian law to assess the reliability of large language models (LLMs) on national legal systems outside the U.S. It evaluates 15 contemporary proprietary and open‑weight models, finding that while closed‑choice accuracy can be high (up to 0.905), free‑text factual correctness never exceeds 0.45. The study reveals a strong dissociation between answer relevance and correctness, systematic variation by legal area, and suggests that expert supervision and grounding in authoritative sources are needed for reliable Colombian legal tasks.
By Rub\'en Manrique, Michelle Castellanos, Jorge Morales, Juan David Guti\'errez, Antonio Barreto Rozo, Joaqu\'in V\'elez Navarro
The paper demonstrates that a natural-language interface can be trained offline to control a xenobot—a synthetic multicellular construct—by using a vision‑language model to judge whether archived intervention outcomes match language descriptions. By treating existing intervention–outcome data as a fixed dataset, the authors train a language‑to‑intervention mapping without new experiments, achieving 80% accuracy on held‑out data compared to a 66.7% baseline. This approach shows that language‑driven control of living systems can be learned purely from archival data and automated visual assessment.
By Nam H. Le, Douglas Blackiston, Michael Levin, Josh Bongard
The paper introduces a method for looped transformers that share a key‑value cache across recursions, reducing memory usage without sacrificing quality. Experiments show that the Looped Prediction Transformer (LPT) and its hybrid variant achieve lower perplexity on FineWeb‑Edu while using 76‑79% less context memory compared to standard transformers. Analysis reveals that shared memory develops distinct representations and serves as a gradient highway, enabling later recursions to focus on shared information.
By Giovanni Monea, Keshav Ramji, Yousef El-Kurdi, Luis A. Lastras, Yoav Artzi, Nathan Godey, Ram\'on Fernandez Astudillo
Inherit-MAS introduces explicit inheritance at both workflow and execution levels for multi‑agent systems built from large language models. A meta‑model generates a workflow of specialized agents, while a judge evaluates each candidate and diagnoses deficiencies. During refinement, workflow inheritance discards unhelpful nodes and applies validated edits, and execution inheritance reuses stored results only when the request and context match, reducing redundant computation.
By Songtao Wei, Yi Li, Zhichun Guo, Bingzhe Li
DAGS introduces a lightweight, attention‑free conditioning scheme that disentangles appearance and geometry for a frozen image diffusion transformer (DiT), enabling high‑fidelity, temporally stable renders with independent control. Two small convolutional encoders generate per‑frame conditioning features, which are injected as learned residuals into the image tokens, avoiding the quadratic cost of attention. Coupled with a recurrent lighting stabilizer and a training‑free temporal guidance term, DAGS transforms a per‑frame image model into a streaming renderer that outperforms real‑time denoisers and diffusion renderers in PSNR and temporal stability while requiring far less compute than path tracing.
By Karthik Mohan Kumar, Damian Andrysiak, Pedro Antonio Pena, Kunal Tyagi, Rama Harihara
Pincer is a defense mechanism for coding agents that operates at the resource layer, complementing existing tool‑call defenses. It uses a digital twin—an isolated‑context model that learns and enforces dynamic, user‑specific least‑privilege policies—to act as a proxy for the agent’s permission requests. The authors introduce a user‑centric dataset to emulate the learning phase and demonstrate that Pincer outperforms several baselines in both security and utility, especially against adversarial attack types.
By Mayank Rathee, Alexander Stepanov, Shalin Madabhavi, Jinhao Zhu, Raluca Ada Popa, Ion Stoica
The paper introduces CausalBridge, a framework that uses causal structure to assign semantics to unnamed variables in numerical measurements. By discovering a causal graph from the data and solving for variable embeddings constrained by this graph, the method aligns variable meanings with a language model, outperforming association‑based approaches. Experiments on questionnaires and robotics scenarios show high accuracy even when most variable names are masked, enabling rapid and cost‑effective system naming.
By Shuhao Zhang, Xuran Zhou, Han Guo, Pengtao Xie, Yujia Zheng
TPBench is a benchmark for dialogue compression that focuses on turning-point eviction, where a user’s change in intent or constraints is lost during compression. It evaluates three information targets—initial user goal, current slot value after revision, and both together—using existing dialogue-state annotations from MultiWOZ and SGD, without requiring new crowdsourcing. The benchmark shows that different compression methods rank differently across probes, and that deleting the turn containing a user update significantly harms current-value accuracy, while removing irrelevant turns does not.
By Minji Park, Seunghyun Yoon, Hyuk Lim
Dynamic Expert Pruning for Multi-Agent Systems introduces DEP, a method that generates a per-request mask of experts based on an agent’s system and task prompts. Unlike static pruning, DEP uses a lightweight predictor trained on workflow transcripts to activate only the experts needed for each specific request, requiring no per-configuration calibration. Experiments across various tasks, models, and MoE architectures show DEP outperforms static pruning and merging baselines, especially when few experts are retained, and it generalizes to unseen workflows.
By Jabin Koo, Soheil Abbasloo, Sungjae Lee, Jungseul Ok