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.
Learning to Zoom Efficiently with a Contrastive Curriculum
Falko Helm, Iryna Gurevych
SFT-free RL teaches MLLMs zoom-in via an InfoNCE hard-negative reward; convincing, honest work, but gains modest and fragile across base models.
MemoryLACE: Memory Lifecycle-Aware Consolidation and Evidence Retrieval
Meriem Yacoubi, Pia Schmidt, Nenad Petrovic, Ahmed Frikha et al.
Proposes sparse lifecycle relations for agent memory; strongest same-backbone results and 2.99x speedup, but key margins are small and lack significance tests.
From Trait Vectors to Circuits: Tracing Refusal and Sycophancy Through Language Models
Oscar Miró López-Feliu, Maya Ozbayoglu
Asks whether steering circuits equal natural circuits: refusal yes (compact), sycophancy partial; rigorous single-model EAP-IG study, no code release stated.
Generative Nested Sampling of Atomistic Thermodynamic Landscapes
Alessandro Coretti, Nico Unglert, Sebastian Falkner, Georg K. H. Madsen et al.
Conditional flows replace MCMC in atomistic nested sampling: >100x fewer energy evals, ~1/3 faster, but demonstrated only on an 8-particle 2D LJ system.
Coupled Tensor-Tensor Completion Method with Applications in Drug Repurposing
Maryam Bagherian, Albert Hung, Ivo Dinov, Joshua Welch
CTTC extends tensor completion to tensor-valued side information with big RSE gains on DTD/LINCS; compelling, though the runtime claim conflicts with Table 2.
Do LLMs Choose Like Humans? Using Cognitive Theory to Evaluate LLM Decision-Making
Johnathan Sun, Andrei Shleifer, Yonatan Belinkov
12 LLMs show human-like context shifts in choices but not feature reweighting (only 26% human-like) on a 140k-trial benchmark; rigorous, theory-driven.
Portable Causal Fairness Across Synthetic Data Generator Families
Steven Golob, Sikha Pentyala, Martine De Cock
Edge-cut fairness is causal factorisation's, not DECAF's — shown on 9 generators, 3 families; solid, but 'fairest' diffusion leans on DemPar utility collapse.
Jina-OCR-v1: Efficient Document Parsing with Speculative Decoding and Dense Verifiable Rewards
Alejandro Barón García, Feng Wang, Emilia Garcia Casademont, Han Xiao
Jina-OCR-v1 pairs DeepSeek-OCR with FastMTP drafting and GRPO under graded verifiable rewards; honest benchmarking, but the recipe's key ablations are missing.
Reducing Catastrophic Risk from AI with Systematic Monitoring and Evaluation of Rogue AI Progression
T. Bauer, W. P. Kegelmeyer, E. Begoli, A. Sadovnik et al.
A cross-FFRDC workshop paper proposes a 5-stage observable/indicator framework for rogue-AI risk; useful scaffolding, but threshold-free and unvalidated.
Improving global precipitation forecasts with an AI weather model trained on satellite observations
Julian F. Schmitt, Bertrand Delorme, Robert C. King, Yashica Patodia et al.
Retraining AIFS on IMERG satellite precipitation (Laxmi) cuts CRPS 19% and drizzle bias 33%, a strong but IMERG-verified result needing independent ground truth.
LLMs Learn Better In-Context from Rules than from Examples
Xiang Fu, Seungmin Cho, Yukyung Lee, Najoung Kim
Claims rules beat examples for in-context learning; a controlled 5-task, 3-family study largely supports this, with idealized-rule caveats.
The 2026 PNPL Competition: Word Classification and Efficient Cross-Subject Generalisation in LibriBrain100
Francesco Mantegna, Gereon Elvers, Dulhan Jayalath, Gilad Landau et al.
Competition paper for 2026 PNPL word classification on LibriBrain100; strong benchmark design and infrastructure, though only early reference baselines exist.
MasterControl Seventeen Every Time
MasterControl AI Lab
Governed analytics: model interprets, policy executes approved programs (completeness, replay proofs); 8B planners 0/330 vs 110/110 policy — narrow benchmark.
VoxReason: Auditing Source-Grounded Speech Plans Before Synthesis
Mengzhe Geng
Proposes a 100-case benchmark auditing speech-plan grounding via cue edits and a deterministic verifier; solid controls, but no learned model is tested.
FlowBalance: Verifier-Grounded Self-Improvement from On-Policy Reasoning Experience
Zixun Huang, Kishan Panaganti, Haitao Mi, Leowei Liang
FlowBalance: verifier-gated dense self-guidance inside trajectory balance; consistent math gains over FlowRL/GRPO but small margins, no significance tests.
The Analyst in the Prompt: Role, Retrieval, and Memory Biases in LLM Financial Analysis
Ahmed Asaad, Amr Mohamed, Yang Zhang, Omneya Abdelsalam
A careful 12-LLM audit on 3,575 SEC filings finds role prompts shift supposedly neutral evidence scores; convincing, with causal and external-validity gaps.
Selective Hypergraph Refinement for Frozen Graph Clustering
Zimo Si
Label-free hypergraph post-processing for frozen graph clustering; gains tiny (0.07-0.14 pp) but the evaluation is unusually honest and rigorous.
SWIM: Student Writing Simulation via Proficiency-Conditioned Generation
Heejin Do, Jakub Kontak, Mrinmaya Sachan
Claims LLMs can simulate student essays conditioned on trait scores: prompting weak, SFT strong, AES-rewarded RL best; credible but rests on an AES proxy.
What is Smoothness?
Zachary P Bradshaw
Cayley-Laplacian irrep ordering ω(σ)=1−Tr[Mσ]/dσ, cleanly axiomatized; math checks out but core is admitted-standard and ML claims untested.
Assessing Adversarial Robustness of Latent Reasoning Models
Shaolong Chen, Ang Li, Mingjie Li, Yisen Wang
Claims latent-reasoning models are less adversarially robust than explicit CoT; a credible, controlled study — small samples and mixed backbones temper it.
Counterfactual Fairness Audits of Multi-Step Clinical LLM Agents Require a Measured Per-Action Instability Floor
Rohith Reddy Bellibatlu, Manpreet Singh, Deepak Parashar, Rahul Joshi
Claims fairness audits need a measured per-action instability floor; rigorous, unusually candid, but only 16 vignettes, 2 models, no disparity results yet.
Swiss-Knife: A Framework for Reconfigurable Externalised Multi-Objective Alignment at Decode Time
Agnibh Karmakar, Mayur Parvatikar, Shreyash Dhoot, Amit Dhanda et al.
Decode-time multi-objective alignment framework with a representation theorem and predictive ablations; gains real but modest, judge-dependent, ~11.5x slower.
Speculative Macro Commit for Faster Tool-Using Agents
Zeyu Liu, Souvik Kundu, Peter A. Beerel
SMC commits verified multi-step draft work to cut tool-agent latency (~10-19% vs SA; ~45% vs sequential on AppWorld), but evaluation is one model pair, statistically untested, and cost-blind.
Language-encoded network topology enables large language models to reason about complex networks
Ucchwas Talukder Utsha, Sakib Mostafa, James Zou, Md Tauhidul Islam
BioGlyph encodes topology as named structural roles, boosting frozen 8B LLM graph QA; rigorous study, but gains fade on frontier models and sparse graphs.