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.
From Collaboration to Capability: Internalizing Routed LLM Experts into Compact Reasoners
Frank Nie, Shuyao Wang, Ethan B. Liu
Claims expert-augmented RL + trajectory SFT internalizes routed experts into compact reasoners; gains credible but single-seed, no significance tests.
Quality-Constrained Routing over a Fixed Pool of Quantized Mixture-of-Experts Instances
Zhenghong Huang, Hongfan Wu, Jiheng Zhang
Rigorously scoped routing over fixed quantized MoE instances; theory and honesty are strong, but the 2.5% FWP gain on 88 prompts lacks variance estimates.
Meddies-PII: A Multilingual Framework for Personally Identifiable Information Extraction in Clinical De-identification
Linh Uyen Le, Christian Hoang, Huy Hoang Ha
Synthetic 1M-doc, 17-language clinical PII dataset; its model tops 16 benchmarks (0.827 vs 0.658 mean F1) but ablations and real-record validation are absent.
Clustering-Based Balanced Sampling and Allocation with Data Parallelism for High-Performance Fine-Tuning
Hyunjin Kim, Youngeun Nam, Jaemin Han, Wonhyeok Choi et al.
Gradient-cluster balanced sampling for DP fine-tuning cuts training time up to 69.6% at near-parity; credible method, but gains are modest and under-tested.
SCOPE-OPSD: Fisher-Conditioned Privileged Subspaces for On-Policy Self-Distillation
Yunmeng Chen, Kunyu Wang, Peihan Li, Yi Wang et al.
SCOPE-OPSD adds a Fisher-conditioned rank-64 hidden-residual loss to OPSD, beating matched random in 10/12 runs; results plausible but lack significance tests.
Distortion of AI Alignment Revisited: RLHF is a Decent Utilitarian Aligner
Kazusato Oko, Annie Ulichney, Nika Haghtalab, Han Bao
Theory paper proving RLHF's exponential distortion stems from data/reference-policy mismatch, with tight Θ̃(βB+β) bounds; main caveat: loose constants.
Correlation-Guided Fast Machine Unlearning via Hessian Analysis
Ayushi Thakur, Ruchir Gupta, Amit Kumar Jaiswal, Prayag Tiwari
Closed-form Sherman-Morrison shortcut speeds unlearning of correlated points; sound algebra, but MIA 0.660 is spun as success though worst of the three.
TokenMapper: A Step Toward Interoperable Speech Token Translation
Tal Kozakov, Tal Rosenwein, Eliya Nachmani
Claims feasible direct speech-token translation across 3 codecs (WER 5.85-9.98% vs native 3.29-4.75%); results plausible but lack stats, baselines, code.
Tight Sampling Complexity with stochastic gradient oracles in Fixed Dimensions
Weiming Ou, Xiao Wang
Proves tight Θ(log(1+κ)+σ²/(με)) query complexity for fixed-d log-concave sampling under noisy gradients; strong theory, AI-co-built proofs need checking.
Reproducing and Evaluating the Generalizability of Subliminal Learning in Open-Weight Models
Daan van der Weijden, Nathan Brack, Selene Baez Santamaria
Open-weight reproduction confirms subliminal learning; extensions show effects vary by trait, task, model. Convincing, though inferential statistics are thin.
Where Decoder Cosine Similarity Fails for SAE Feature Flow Discovery
Hendrik Droste, Christian Medeiros Adriano, Kathrin Korte, Holger Giese
Claims pairwise SAE decoder cosine misses most ablation-validated feature transitions (88% Pythia, 54% Gemma subset); solid causal method, thin generality.
SIMS: Scale-Invariant Merit-Function-Based Scalarization for Multi-Task Learning
Zebin Chen, Fei Xing, Yang Chen, Hua Liu et al.
SIMS proves a log transform uniquely yields scale-invariant merit scalarization for MTL; clean theory, but benchmark gains are small and lack seed variance.
ResoSeg: Resonance Tagger using Transformer and Segment Model
Chunkai Li, Junhao Yin, Ke Li, Jingde Chen
First segmentation-based resonance tagger for BESIII ηc (ParT+SAM), ~2.6× conventional efficiency in MC per authors; MC-only, no SPANet baseline.
SteerDuplex: Steerable Duplex Speech Dialogue Models
Utkarsh Tyagi, Ramaneswaran Selvakumar, Advait Gosai, Sonal Kumar et al.
SFT+RL lifts Moshi duplex speech steerability 44.5pp on new SteerBench; substantial, honest gains, but thin baselines and single-judge scoring limit certainty.
Geometric-to-Semantic Spherical Transfer Learning for Cortical Sulci Labeling
Saeb Tounsi, Joël Chavas, Pietro Gori, Vincent Frouin et al.
Promising transfer learning for sulci labeling: SSL on 30k UKB + soft-init TPI lifts Dice to 0.77 vs 0.72 from-scratch; small N=22 test, modest margin.
SIFPBPNet: A Dual-Path Network for Wearable and Cuffless Blood Pressure Estimation via Individualized Steady-state Representation
Shuailong Tang, Xiaoyu Li, Donglin Xie, Wei Chen et al.
Claims a GAT steady-state prior boosts cuffless BP accuracy (SBP MAE 8.57 mmHg); promising but gains may reflect history access and lack significance tests.
The Agentic Company OS: Substrate Inversion for Sustained Enterprise Agent Deployment
Oliver Aleksander Larsen, Mahyar T. Moghaddam
Position paper: enterprise agent stalls stem from human-first data substrates; rebuild them as Markdown for LLMs. Lucid, honest, but wholly unvalidated.
SWARM: A Multilingual Human-Annotated Dataset for Russian Propaganda Detection in Search Engine Results
Manuel Tonneau, Abhinav Dubey, Farhan Shaikh, Ilaria Vitulano et al.
SWARM: 2,183 search results in 9 languages labeled for Russian propaganda-narrative support; blocklists fail (F1 0.39); best LLM 0.73, still unreliable.
LifeMem: Enabling Lifelong Experience Reuse for LLM Agents
Yuli Qiu, Yutong Li, Wei Su, Zeming Liu et al.
LifeMem clusters trajectories into workflows to distill reusable agent skills; broad 10-env study supports gains, but no significance tests or variance given.
LiveProBench: Can Streaming Video Models Really Interact Like Humans?
Kaixuan Du, Xin Wan, Hang Zhang, Meng Cao et al.
Solid streaming benchmark for proactive interaction with paired silence scoring; a human-score inconsistency, small subtasks, and no data link temper it.
Doc2FRC: Length-Consistent Document-Level Machine Translation via Fixed-Range Chunking
Xiaotian Wang, Youyuan Lin, Zhan Shen, Hitomi Yanaka
Proposes DP fixed-range chunking to align train/test chunk lengths for 7B DocMT; promising IWSLT gains, but edge over simpler chunking baselines is modest.
Separating Engineering Reasoning from DEXPI Serialization in LLM-Based Greenfield Surface-Process Design: A Three-Case Study for Underground Gas Storage
Qingchuan Zhu, Shuyue Tong, Pengju Ren
Deferring DEXPI serialization cut prompt tokens 174-197x and isolated failure modes, but n=1 runs per condition and confounded bundles limit this diagnostic.
Hardware-Aware Learned Representation Compression for Distributed In-Sensor Vision
Chengwei Zhou, Abu Masum, Xuming Chen, Mehran Moghadam et al.
OASIS claims up to 18,816× in-sensor communication cut and 2–4.5× energy savings with FPGA-grounded co-design; credible but headline ratios mix configs.
Generative AI Use Cases In Real Estate Marketing: Adoption and Constraints in Germany
Victor Kolominsky-Rabas, Leopold Müller, Felicia Perpina, Niklas Kühl
Eleven-interview study maps current vs potential GenAI use cases in German real estate marketing; credible qualitative work, limited by AI-user-only sample.