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.
Stochastic Reconfiguration as Statistical Filtering for Overparameterized Neural Quantum States
Tak Hur
Reframes SR's diagonal shift as ridge-regression regularization plus MS-SR; ablations support it, but online energy gains are not statistically significant.
Human-guided physics-constrained AI agents construct an auditable model of soil-plug evolution
Jie Shi, Yimin Lu, Zhongkun Ouyang
Claims a human-governed multi-agent workflow built an auditable soil-plug solver (12.4%/4.2% MAPE); credible, well audited, but in-sample on a tiny benchmark.
You've Seen Enough: Quality-Constrained Image Coding for Machines
Khoa Pham-Dinh, Sanaz Nami, Hamed Rezazadegan Tavakoli, Moncef Gabbouj et al.
Constrained ICM with bilinear penalty gives solid Cityscapes BD-rate gains (-29.8% vs baseline); oracle lambda choice, single dataset, no stats temper it.
LLM-Based FORM Code Generation with Verification-Driven Fine-Tuning
Bakar Chargeishvili
宣称8B模型QLoRA微调后超越756B前沿模型生成FORM代码；B1/B2证据扎实且报告诚实，但小样本开放式基准尚无定论。
What Can a Recurrent State Safely Forget?
Linzhe Zhang, Changming Xu
Formalizes which RNN hidden-state directions are safely forgettable via a predictive quotient and PAC audits; sound theory, small-scale experiments.
TicTacBench: Benchmarking Timing Closure Capabilities of Coding Agents
Bowei Wang, Zhigang Fang, Zhijie Yang, Renzhi Chen et al.
TicTacBench adds 30 post-PnR RTL timing-closure tasks; across 300+ runs of 8 LLM agents the best closes 53.3% — a credible open benchmark, but rankings rest on single untested runs.
Machine-Interpretable Information: Compiling Documents into Searchable and Readable Protocol States
Yifan Wang, Dejing Dou
Compiles docs into 56-token states readable by frozen LLMs; Residual-MII beats full-context EM on HotpotQA at ~7% FLOPs, but F1 lags and evidence is thin.
The Evidence Ladder for Reinforcement Learning in Healthcare: From Retrospective Policies to Trusted Interventions
Yunfan Zhao
Narrative survey of healthcare RL via a six-rung evidence ladder plus an evidence card; conceptually strong but methodologically thin and unvalidated.
One to More, More to One: Category-Aware Iterative Expert Training for Software Engineering Agents
Jie Zhao, Ziyu Jiang, Suhang Zheng, Minghui Shan et al.
Alibaba trains three category RL experts for SWE agents and merges them via multi-teacher distillation; gains look real but a compute-matched single-policy baseline is missing.
If You Hear It, Help Find It: Orthogonal Knowledge Distillation for Open-Vocabulary Audio-Visual Event Localization
Yi Xu, Cheng Chen, Wenzhuo Lei
Proposes reliability-aware asymmetric distillation (OV-OrthKD) for open-vocabulary audio-visual localization; well-ablated 0.816 AP gains, but evidence rests mainly on one benchmark.
Leaky-integrator reconstruction: taming error accumulation in recursive differenced time-series forecasting
Zijiang Yang
Recursive differenced rollouts diverge via unit-pole integration; a training-free leaky integrator cuts H=336 error 43–75%. Convincing but small-scale.
Discovering Physical Representation Languages
Linzhe Zhang, Changming Xu
Claims provable recovery of a discrete physics ontology from anonymous experiments; rigorous theory and honest audits, but evidence remains largely synthetic.
Bayesian Filtering in Physical Systems via Test-time Trained Flow Matching
Ruiqi Feng, Chongyi Wang, Tao Zhang, Tailin Wu
BFF encodes filtering posteriors in flow-matching weights via test-time training; strong PDE/tokamak results, but particle baselines and error bars are thin.
Blind Thermodynamic Ontology Discovery from Anonymous Experiments
Linzhe Zhang, Changming Xu
Claims thermodynamic coordinates are identifiable up to contact gauge via four experiments; theory elegant, results strong but low-noise and baseline-light.
Leveraging Industrial Foundation Models at the Edge of Particle Physics Detectors via Distillation Learning and Hardware Co-design
Gia Ancone, Qibin Liu, Liangyu Wu, Julia Gonski
First TimesFM fine-tuning for HEP DAQ, distilled to 25 ns FPGA models; teacher results strong, but compressed-model benchmark wins are not clearly shown.
Enhancing Shrimp Disease Detection via Deep Learning and Data Refinement for Resilient Aquaculture
Vinh Canh-Thanh Truong, Hai-Binh Pham, Ngoc Hong Tran
Claims first ViT+SimCLR for shrimp disease: supervised ~96% val vs 85% SSL; promising baseline, but 70-image test, no stats, and abstract/results mismatch.
Tool-Augmented On-Policy Distillation for LLM Domain Adaptation in Sequence-Based Omics Tasks
Jie Ying, Zhefan Wang, Zihong Chen, Zhengqing Li et al.
First omics reasoning benchmark; TA-OPD lifts prediction and evidence grounding. Strong releases, but LLM-judge circularity and small-n stats temper claims.
RiverVLN: Phase-Grounded Temporal Vision--Language Navigation for Unmanned Surface Vehicles
Jieling Wu, Yuehao Huang, Jiajun Lv, Tao Huang et al.
Claims first USV river VLN benchmark plus a phase-grounded policy at 0.79 SR in sim; honest about drift, but baselines are weak and statistics thin.
OmniEcho: Audio-Visual Spatial Understanding for Omni-Modal Embodied Agents
Ruixun Liu, Yuxuan Wang, Jiacheng Xie, Yuhuan You et al.
Introduces a real-world FOA benchmark (2,972 QA + 900 nav) and an FOA-augmented omni model; gains are real but modest, SOTA claims rest on its own benchmark.
MuLA-Bench: A Multilingual Long-Form Audio Understanding Benchmark via Multi-Tier Auditing
Zeyu Yang, Xinyu Zhang, Zibo Bi, Pei Zhang et al.
记录详实的 16 种语言长音频基准测试；条件难度的研究结果可信，但基于 LLM 构建的黄金标准（gold）和上下文限制影响了模型间的对比。
Beyond PUE: A Local Impact Audit Framework for Data Center Environmental Accountability
Sharifa Sultana, Syed Ishtiaque Ahmed
Proposes a five-category local-impact audit framework; diagnosis of location-blind metrics is compelling, but the framework is never numerically demonstrated.
TRACE: Tractable Routing Autoencoder for Clinical ECG
Shunbo Jia, Runze Ma, Haonan Lyu, Haijin Zhang et al.
Claims knowledge-specified routing beats ECG foundation models at 1/8 params; own controls show ontology adds no F1 and pathway interventions are non-selective.
WaveletECO: A Closed-Loop Physical ECO Platform and a Specialized Local Language Model
Guoxiang Xu, Guozhen Ji, Zijian Luo, Zhengrui Chen et al.
Trained local 9B model beats GPT-6 Astra on 22-design ECO benchmark at ~1/147 cost; plausible, but zero-SD repeats and seen-design overlap need scrutiny.
RLVR$^{2}$: Reinforcement Learning with Verifiable Rubric-based Ranking
Hao Li, Zhengkun Zhang, Gangqiang Hu, Zhen Zhang et al.
RLVR² turns rubric scores into within-group Bradley-Terry rankings for RL reward; broad gains at 3 scales, but single-run, no significance tests, closed data.