Evolve Quick Reviews
Public Paper Reviews
Structured preliminary reviews of recent open-access papers, generated and continuously updated by the Evolve analysis engine. Each review condenses what the paper claims, how solid the evidence is, and the questions a rigorous peer reviewer would ask — English analysis with abstracts in your language. These are independent AI analyses, not peer review; the canonical record always lives at the publisher.
CoVeR: Coverage-Based Routing of Verifier Calls in Agentic Retrieval
Daeyoung Roh, Donghee Han
CoVeR skips 62-68% of verifier calls in agentic retrieval via one coverage threshold at matched EM; rigorous, credible, with cost and weight-release caveats.
SGD-KV: Summarization Guided KV Cache Compression
Zeyu Liu, Woomin Song, Xuandi Fu, Sai Muralidhar Jayanthi et al.
KV budget by summarization-head score cuts cache 75% with solid 1M-token results; train/eval overlap and no variance stats temper the SOTA claim.
Trust Me, I'm Your Developer: Self-Issued Authentication in Large Language Models
Syed Ghazanfar Abbas, Dongyan Xu
Coins MIPC/CFA: Qwen and Mistral 'Verified' a developer claim via a self-made quiz. Sound framing, but single-trial, unversioned models, no transcripts.
What Else Needs Fixing? Exploring Cost-Effective Test-Time Compute for Revision Propagation in Artifacts Generated Through Conversation
Daisuke Kikuta
New benchmark for LLM revision propagation in conversational JSON artifacts; solid cost-aware evaluation, but small synthetic data with single-annotator gold.
B2B Customer Conversion Prediction: A Document Representation, Graph Theory, and CatBoost Driven Methodology
Tianqi Wang, Sheikh Shams Azam, Wan Eih Huang, Anton Wiranata et al.
Claims a B2B conversion pipeline reaching 91% accuracy with CatBoost on proprietary HP data; solid engineering, but weak evaluation and no external validation.
MedQA-MM: Shortcuts Behind Medical Visual Reasoning
Benlu Wang, Yifan Zhang, Jiaqing Yu, Chin Siang Ong et al.
Audits six medical MCQ benchmarks for shortcut cues inflating VLM scores; strong matched-repair evidence, but repairs are GPT-authored; expert review sampled.
Contextual Tamil Spelling and Grammar Correction Using Progressively Fine-Tuned Sequence-to-Sequence Transformers
Karthikeyan A, Jaya Nirmala S, Sangeetha Sivanesan, Indhu R et al.
Fine-tunes mT5/mBART-50 on synthetic Tamil pairs; best 69.3% top-1 with clean staged ablations, but all data synthetic and no code release limits impact.
Guiding the coarse levels of semantic IDs makes the fine levels learnable
Bin Wang, Zhengyu Zhang
Guided SID pins coarse Semantic ID levels to a task attribute; matched A/B retrieval gains on Meta ads data, but evidence is early with an unverified citation.
Risk and Anomaly Identification for Distribution Network Optimal Operation Based on Reinforcement Learning and Uncertainty Quantification
Ziqi Zhang
Uncertainty-aware DRL splits aleatoric/epistemic uncertainty for grid risk/anomaly ID with MISOCP fallback; theory rigorous, but 62–76% fallback and no code.
Less Is Moral: A CHARMing Framework for Moral Foundations Detection in Endorsement Behaviour
Huixiang Fu, Marian-Andrei Rizoiu
CHARM: theory-grounded moral detector on LoRA LLaMA-8B with strong OOD gains; internal inconsistencies and dictionary-derived MAC supervision temper claims.
Time Without Timesteps: Simulating Coupled Dynamical Systems via Self-Consistency
Liyu Zerihun, Mark Shinyoung Lee
Waveform relaxation with learned trajectory surrogates turns coupled ODE simulation into fixed points; a careful, honest early exploration on small systems.
PACE: Towards Surfacing Hidden Conflicts in User Requests
Yoojin Kim, Jihyoung Jang, Hyounghun Kim
PACE 在个性化 KB 中对隐藏冲突进行了基准测试，PACEMAKER 提升了检索效果；但相同模型的生成与评估，以及缺乏显著性检验，削弱了其适度提升的可信度。
Multilingual Agent System for Inclusive Wildfire Evacuation Guidance
Shruti Kulkarni, Lynn Tong, Aditi Namboodiripad, Chelyah Miller et al.
BEACON: sound multilingual wildfire evacuation system design, but all evidence is qualitative—no metrics, baselines, or user studies back the claims.
DE-Venus: A Data-Efficient RLVR Framework for Large Language Models
Shenzhi Yang, Guangcheng Zhu, Kai Tang, Zhengqing Zang et al.
Unified verl-based framework for data-efficient RLVR; well-engineered with real savings, but small margins lack variance checks and key methods are self-cited.
Latent Energy Action Planning with World Models
Phu Pham, Aniket Bera
LEAP replaces CEM with gradient dual-energy planning through a frozen world model, lifting 4-task mean success from 77.5% to 94.8%; rigorous but narrow baselines.
Decoupling Turn-Taking from Semantics: A Decoupled Data Approach for Finite-State-Machine-Based Full-Duplex Dialogue
Yihang Li, Chenhui Chu
Proposes decoupled HH/HA data with rule-based FSM tapes and SAC loss; strong gains over synthetic baseline, but turn-taking evaluation is partly circular.
Geometry-Aware Graph Construction via Adaptive Spectral Bandwidth Control
Ecem Bozkurt, Antonio Ortega
Per-node bandwidth sets kernel effective rank to MST intrinsic dimension; consistent CIFAR-100 gains, but one dataset and no adaptive baselines compared.
Feature Reconfiguration With Visual Prior for Medical Lesion Segmentation
Yinan Liu, Jiankang Hong, Zhen Gao, Ye Luo
FreNet reconfigures features with SAM priors (IPNN+DFR), gaining broadly on 9 datasets, but its Table 2 shows a CVC-300 drop vs SAM; SOTA claim needs nuance.
When Seeing Overrides Knowing: Visual Dominance and Deferral-Based Method for Personalized Safety in VLMs
Edward Sun, Yuchen Wu, Zixian Ma, Eric Hanchen Jiang et al.
Introduces MPS-Bench, a visual-dominance mechanism, and PRISM deferral monitor (0.978 AUC); strong, but label circularity and judge overlap need scrutiny.
Gradients Know What Outcomes Don't: Unlocking Reinforcement Learning for LLM Reasoning with Gradient-Aligned Rewards
Leqi Zheng, Jinbo Su, Fang Niu, Chaokun Wang et al.
Paper claims expert-CoT gradient-cosine rewards boost RLVR math pass@k at <9% overhead; well-motivated with theory, but gains are small on tiny benchmarks.
Beyond .WAV: Design and Software Verification of VocalCap, a Traceable Browser-Based Audio Capture System for Vocal Biomarker Research
Augusto Camargo
VocalCap adds traceable browser voice capture; verification is rigorous, honestly scoped, yet self-conducted: synthetic E2E plus a 39-file internal pilot only.
How Perturbations Propagate: A Multi-Level Analysis of Robustness in Large Language Models
Dun Li Chan, Emily Liu, Niyathi Allu, Christian Hoang
声称六种扰动类型在 GPT-2/Qwen2.5 中留下了可区分的多层级特征；这是一项可信且客观的研究，但纯粹依赖描述性统计削弱了其说服力。
CNA: An AI-Oriented Comprehensive Normalized Assessment for Healthy Status and Application to Optimize RRT Strategies by Reinforcement Learning
Jiang Liu, Chan Zhou, Yujie Li, Di Wu et al.
CNA vital-sign score guides offline RL for RRT; retrospective mortality 13.2%→5.0% gain is promising but confounded by indication and self-defined evaluation.
EvoHarnessBench: Can Your Agents Keep Pace with an Evolving Harness?
Zixuan Ke, Vaidehi Patil, Haizhou Shi, Yang Li et al.
针对 LLM 智能体工具演进的首个基准测试；设计扎实且具备机制分析，但启发式的技能/智能体标签和单薄的数据统计削弱了其结论的说服力。