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.
How Lossless Is Lossless Speculative Decoding? The Role of Numerical Precision in Orthrus
Ilya Koziev, Leonid Sinev, Ivan Oseledets
Independent reproduction: Orthrus 'lossless' decoding matches AR output only ~45% under BF16 but 100% under FP32; credible, though FP32 exactness unexplained.
Psychosis involves a deficit of information compression in connected speech
Samuele Vallisa, Claudio Palominos, Rui He, Emre Bora et al.
Finds reduced grammar-based surprisal gain and embedding ID in 106 Turkish psychosis patients; credible, but ID-surprisal effects flip sign by estimator.
Automating Attack Graph Construction for Agentic Pentesting. Towards Neuro-Symbolic Vulnerability Hunting
Oliver Stevanovic, Jasmin Wachter
A semi-automated scanner-to-MulVAL pipeline shows feasibility on 54 web CTFs (100% goal reach, 53.7% coverage, 83.9% noise) but no agent-level evidence yet.
To Each Language Its Tokenizer: Modular Tokenizers for Efficient Multilingual LLMs
Franck Signe, Hippolyte Pilchen, François Yvon, Édouard Grave
Modular tokenizers give each language a monolingual-grade subtokenizer and cut inference cost; compression evidence strong, downstream parity single-run only.
The Misery of Mechanistic Interpretability: A Formal Perspective
Tobias Ladner, Matthias Althoff
First formal per-layer faithfulness certificates for LLM SAEs/transcoders; sound and novel, but local guarantees only and certified overlap stays low.
Strong and Compact Policies for Submodular Markov Decision Processes via LP-Based Submodular Orienteering
Lars Rohwedder, Rico Zenklusen
LP-based Sherali-Adams rounding gives O(log H)-approx Submodular MDP policies (first o(H)); rigorous theory, but purely theoretical with no experiments.
EEG-Xplain: Decoding Neural Black-Boxes of EEG Foundation Models
Hansong Ma, Junxiao Wang
A unified XAI framework interpreting EEG foundation models; spatial fidelity is strong, but TP-only sampling, small n, and unvalidated LLM reports weaken it.
The Token Before the Value Is the Key: How Hybrid Architectures Organize Induction Circuits
Ke Cheng, Xin Xu, Yixiao Chen, Lei Xin et al.
Claims hybrids put induction Carrying in efficient layers and Matching in global receivers, keyed to lag one; patching-probe evidence is unusually rigorous.
Specifying Reward Functions for RL Without Environment Sampling
Stephane Hatgis-Kessell, W. Bradley Knox, Emma Brunskill
EARS learns rewards from preferences over imagined trajectories, zero env sampling; beats LLM prompting often, but tests only near-linear ground-truth rewards.
Bayesian Optimisation Using Product-of-Experts Gaussian Process Models with Uncertainty Calibration
Yean Hoon Ong
BO-pro-c embeds a calibrated product-of-experts GP in BO: synthetic-only tests show real ~39% overhead cut vs a global GP but tiny, statistically unsupported regret gains.
Benchmarking Intra-Patient 3D Deformable Multimodal Image Registration
Matteo Barbieri, Giammarco La Barbera, Juan Pablo De La Plata, Sabine Sarnacki et al.
Benchmarks 9 multimodal 3D registration methods on 3 datasets: no robust winner, metrics disagree; credible and useful but lacks statistical testing.
Assembling the CREW: A Collaborative Multi-agent Reinforcement Learning Framework for Automated Related Work Generation
Hai-Dang Dang, Bao-Yen Pham, Bao Nguyen, Tran Thi Huong et al.
CREW trains LLM agents via IPPO to coordinate related-work writing; citation gains strong but quality margins (+0.06) marginal, no significance tests.
Projection-Free Multi-level Algorithms for Stochastic Constrained Compositional Optimization
Wei Jiang, Sifan Yang, Wenhao Yang, Yibo Wang et al.
Comprehensive projection-free multi-level framework: first FW-gap rates and ε^-1.5 SFO; theory solid, but key baseline LiNASA+ICG missing from experiments.
FairLint-DL: An IDE-Native Tool for Fairness Debugging of Deep Learning Software
Archit Rathod, Saeid Tizpaz-Niari
VS Code tool packaging established QID fairness methods; solid artifact, honest framing, but proxy-model validity and single-run evaluation limit evidence.
Evaluating Open-Weight E-Commerce Agents with Environment-Grounded Verification
Nimit Shah, Haitz Sáez de Ocáriz Borde
Precommitted-scenario e-commerce eval turns success rates into verifiable failure diagnoses; strong design, but simulator/judge dependence limits conclusions.
Language Orthogonalization for Zero-Shot Cross-Lingual Audio Deepfake Detection
Minu Kim, Ji Sub Um, Hoirin Kim
Closed-form LID-to-S3M ridge map (source bonafide only) cuts cross-lingual EER in all 30 cells; simple and broad, but single dataset, no debiasing baselines.
Inference-Engine Fingerprinting Attacks are Practical: Exploring Model-Driven Environmental Discovery, Exploitation, and Escape
Sarah Radway, Andrew Cheng, Vijay Janapa Reddi, James Mickens
Shows LLMs can fingerprint their inference engine from output tokens and chain CVEs toward bare metal; timely threat model, but final PoC stage emulated.
Don't Count the Edits, Judge by the Outcome Alone: Reward-Based Evaluation for Grammatical Error Correction
Hayeong Ryu, Sunhee Jo, Seunguk Yu, YoungBin Kim
SURE trains a DeBERTa reward model on LLM-judged GEC preferences; SEEDA results are strong, but supervision rests on proprietary LLM judges with limited rigor.
FedLTLib: A Comprehensive Benchmark for Federated Long-Tail Learning
Changkun Lin, Junxiao Wang
Promising FL-LT benchmark unifying 13 algorithms under a decoupled imbalance/skew taxonomy; fixed α=0.5, no error bars, and claim-table gaps temper findings.
Through the Eyes of the Beholder: Biometric and Demographic Conditioning for Multimodal Sexism Detection
Ana-Maria Luisa Mocanu, Sebastian Mocanu, Ciprian-Octavian Truică, Elena-Simona Apostol
EXIST 2026 system conditions multimodal sexism detection on annotator biometrics/demographics; honest null ablation, headline signal claim outruns evidence.
Multi-View Molecular Representation Learning with Hierarchical Graphs and Contextualized Fingerprints
Gwang-Hyeon Yun, Jong-Hoon Park, Bing Hu, Helen Chen et al.
Claims 2.77% ROC-AUC gain from dual graph-fingerprint pretraining; solid ablations, but gains lean on noisy ClinTox, 3 seeds, no significance tests, no code.
IROH: Insightful Ranking Of Humor using Multi-Stage Hybrid Retrieval with Rationale-Distilled LLM Judges for JOKER 2026 Track Task 1 English
Ana-Maria Luisa Mocanu, Sebastian Mocanu, Ciprian-Octavian Truică, Elena-Simona Apostol
Claims 1st at JOKER 2026 Task 1 (0.6347 MAP) with an LLM-judge retrieval pipeline; strong ablations, but test-set tuning and no significance tests.
Beyond AI Literacy: A Structured Review and Exploratory Meta-Analysis of Measures for Competent Generative-AI Use
Daniele Veri'
Reviews 24 studies of GenAI-literacy measures, pooling 3 same-program correlations (r=.055, CI incl. 0); candid but thin evidence, battery unvalidated.
Principal-timestep Restricted Init via Sparse Matrix-decomposition in Flow-matching
Jiayang Gu, Zheng Fang, Lichaun Xiang, Fanghui Liu et al.
PRISM-LoRA claims timestep-restricted sparse init fixes spectral-LoRA failure in flow matching; theory neat, gains real but uneven, stats thin.