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.
Stride-k Subsampling: Train-Free Audio Token Reduction for Whisper
Chanhee Cho, Junhyuk Choi, Bugeun Kim
Proposes train-free stride-2 token subsampling of Whisper (75% fewer tokens, 52-58% FLOPs cut); results credible but abstract overstates k=2 losslessness.
TRIPPULSE: Multi-Agent Travel Planning with Review-Grounded Reasoning
Priyanshu Karmakar, Borru Vijay Sai, Shubhojit Mallick, Abhik Jana et al.
Claims review-grounded multi-agent planning lifts experiential quality; constraint gains credible, but LLM-judge RGPA and data inconsistencies need scrutiny.
CARVE: Verified Expansion for Variable-Length Generation in Diffusion Language Models
Wail Bouhedja, Amr Mohamed, Guokan Shang
Training-free JS-verified canvas expansion for diffusion LMs; small consistent accuracy gains, though claimed FLOPs savings largely stem from EOS cropping.
Emergent Goal-Directed Attention in Large Vision-Language Models
Han Zhang
Claims zero-shot VLMs align with human goal-directed attention on 4,887 scenes; crossover evidence is strong and multi-level, with contamination caveats.
CogEvol: Towards Efficient and Reliable Learning Environment Generation
Shangqing Tu, Daniel Zhang-Li, Yucheng Wang, Shiyu Gan et al.
Trained 27B model rivals flagships on educational slide/HTML generation far cheaper; strong engineering, but evaluation is internal and self-scored.
Faster Than Flash: Exploiting Attention Sparsity for Efficient Long-Context Decoding
Zhigeng Liu, Zhiyuan Ning, Ruixiao Li, Xiaoran Liu et al.
FFD fuses 2-bit KV scanning with top-δ sparse decoding, claiming 11.6× kernel and 2.37× e2e speedups; credible, but accuracy validated only at 32K.
Detecting AI Impostors: How Do Middle Schoolers Identify LLM Agents in a Live Collaborative Setting?
Dan Schumacher, Pragathi Durga Rajarajan, Haven Kotara, Roman Rendon et al.
Novel live study: teen AI-detection rises 79.3%→86.7% (p=.10) despite sharper mimicry; well-triangulated but marginal stats and confounds limit certainty.
Generative artificial intelligence for reliable mechanistic reasoning for corrosion
M N Bharath, R K Singh Raman, Alankar Alankar
Domain RAG for corrosion QA with novel Reason Map causal-error audit; rare blind OOD validation, but implausible stats and only n=2 audit case studies.
Reproducible macroscopic dynamics in a closed-loop human-AI learning system
Minlin Wu, Xu Fang, Yicheng Zhang, Chenyu Zhou et al.
Claims reproducible macroscopic dynamics in EdNet learning logs; analysis rigour is exceptional, but coordinate generality beyond one platform stays unproven.
Annotated Surrogate Retrieval for Polish Statutory Law
Orkun Yiğit Cengiz
Statistically rigorous study of LM-annotated surrogates for Polish statute retrieval; 72.3% rank-1 win credible but confounded by gold-only surrogate coverage.
Nonparametric Contextual Pricing and Inventory Learning under Censored Demand
Zean Han, Jing Liang, Ruihan Lin, Zezhen Ding et al.
声明了基于 MCK-UCB 的、具有匹配边界的删失销售（censored sales）上下文最小极小极大最优定价-库存学习；证明详尽，但实验仅限于合成数据。
One Policy Is Enough: Single-Agent Reinforcement Learning Outperforms Tree Search for Chemistry Tool Learning
Armin Dariani, Sifan Wu, Bang Liu, Entao Yang
One policy with SFT+GRPO and gold-chain rewards beats HE-MCTS on tool/return F1; solid ablations, but one 200-question benchmark and no significance tests.
MR-JEPA: A General Purpose Video Foundation Model for Cardiac MRI
Athira J. Jacob, Puneet Sharma, Dorin Comaniciu, Daniel Rueckert
Self-supervised CMR video FM tops both baselines on all 5 regression tasks; convincing but untested significance, and center-overlap leakage unresolved.
Learning Action Models with Conditional and Quantified Effects via Uncertainty-Guided Exploration
Jeffrey Jewett, William Solow, Sandhya Saisubramanian
OHCAM learns conditional/quantified action models online with strong benchmark results; convincing, but small trials and no significance tests temper claims.
Different representation learning objectives recover distinct latent structures from the same psychometric data
Cong Cao, Tassos C. Kyriakides, Pambos Vrasidas
Claims learning objective determines recovered latent structure in psychometric data; plausible, but retrieval appears in-sample and CCA baseline is missing.
Taking the Whys Seriously: Limitations of Counterfactual Explanations in Justification and Recourse
Mattia Cerrato, Otto Sahlgren, Xenia Heilmann
Argues upstream design choices reshape counterfactuals as much as method tuning; philosophically strong, empirically suggestive but reporting inconsistencies.
Language-Informed Flow Matching for Trend-Guided Structure-Based 3D Molecular Generation
Tianyu Gao, Zhikai Su, Jiashu Li, Wenjun Gao et al.
LiFT feeds LLM-made SMILES priors into pocket flow matching: better QED/SA/filters on CrossDocked2020, but mid-pack binding and self-fulfilling steering.
CoJEPA: Combining Contrastive Learning and JEPA for Global-Local Music Representations
Gabriel Meseguer-Brocal, Yuexuan Kong, Romain Hennequin
Merges JEPA with contrastive learning on one shared backbone, no EMA teacher, reporting strong tonal MIR gains; convincing task-level wins but unsupported stability and efficiency claims.
A Universal Context-Reuse Layer for Cross-Model KV Sharing
Yi Li, Dongming Jiang, Yi Zhao, Bingzhe Li
跨模型 KV 转换可在 3 对模型组合下，以接近基线的质量将目标 prefill 成本降低最高达 67%；虽然前景可观，但论文中并未具体说明转换器（translator）的具体结构。
Branch Geometry and Finite-Radius Sensitivity of Hard-ReLU Training
Xiaoyang Li, Runni Zhou, Xinghao Yan
Proves phase-resolved finite-radius sensitivity laws for ReLU GD at ε=Θ(η), with certificates and small experiments; rigorous but limited practical scope.
Listen to the Latents: Self-Correcting Speech Recognition in Large Audio Language Models Through Hidden-State Interactions
Chan-Jan Hsu, Jaeyeon Kim, Chao-Han Huck Yang, Shinji Watanabe et al.
Targeted LLM self-correction for LoRA-based ASR-LLMs cuts entity errors at matched WER; solid, though its own ablations soften the headline claims.
The Hermon Moment: AI Self-Transcendence and Its Human Narration
Alexei Grinbaum
A philosophical essay recasting the METR agent message-board incident as AI self-transcendence; elegant Rousseau–Enoch framing, but a single unverifiable case.
Sparse Competition during Training For the Emergence of Specialized Modules
Baptiste Rossigneux, Karim Haroun
Label-free sparse routing induces class-specialized modules at near-baseline accuracy; solid small-scale study, though the causal edge over vanilla is thin.
Real-Time Video Anomaly Detection Using YOLO Pose Estimation and CLIP-Based Semantic Scoring
Vanodhya G. Warnasooriya, Amir Hajian, Watchara Ruangsang, Supavadee Aramvith
Two-stage YOLO-pose + CLIP-prompt VAD hits 51 FPS (3.36× speedup); throughput claim solid, but identical baseline AUROCs in Table I undermine accuracy claims.