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.
Replan, Repair, or Edit? A Unified Empirical Evaluation of Travel Agents for Itinerary Revision under Resource Disruptions
Xiaofei Yuan, Yan Zhang, Shaobo Qiao, Huangleshuai He et al.
Compares replanning, HTN repair, and LLM editing on TREK disruptions; a careful unified study, though evidence rests on adapters without statistical tests.
Encoded but Not Decoded: Layer-Localized Evidence for a Three-Level Gap in LLM Syntax
Zhenyan Lu, He Wang, Xiaohui Huang
Claims models encode syntax they fail to deploy; careful, honest three-level design, but 48 items and verb leakage inflate small-model probe ceilings.
Reach or Solve? Attributing Agentic RL Gains with Checkpoint Handoffs
Xuan Liu, Jingbin Qian
Checkpoint handoff splits agentic RL gains into reach and solve; solid protocol, positive interaction in all 5 conditions, but small samples, Qwen-only.
Does per-frame early exit pay? A compute-matched study of dynamic depth for on-device speech enhancement
Clément Laroche, Riccardo Miccini
Careful compute-matched study: early-exit training yields an efficient static model family; per-frame routing adds little. Rigorous, but single-dataset.
A Risk-Adaptive and Evidence-Constrained Framework for Generative AI Feedback in Programming Education
Shihao Wang
Careful student-disjoint pipeline linking risk prediction to gated LLM feedback; honest, modest results, but no evidence feedback helps learners.
Beyond Model Size: Redesigning LiSenNet for embedded speech enhancement
Clément Laroche, Rasmus Kongsgaard Olsson
Redesigns LiSenNet into an int8 NPU model with PESQ 3.01 at 4.83 ms/frame on an MCU; credible on-device work, though gains are small and lack error bars.
Ontology-Mediated Neurosymbolic Constraint Acquisition from Multiple Stakeholders
Stefan Bischof, Juliana Kainz, Danilo Valerio
Position paper proposing ontology-mediated LLM constraint acquisition (CONTO + soft priorities, FLEXI microgrid); articulate and honest, but unbuilt and unevaluated.
When Can Agents Forget Their Reasoning? ICLR for Long-Horizon Agent Context Compression
Mingxuan Wang, Fei Luo, Bo Wang, Guorun Yao et al.
ICLR prunes low-entropy agent reasoning online, cutting tokens and lifting reward, but gains hinge on one domain, delete-all nearly matches, stats thin.
Continual Enterprise World Model Discovery in Dynamic Systems
Shambhavi Mishra, David Vazquez, Perouz Taslakian, Marco Pedersoli et al.
Claims a ServiceNow benchmark and agent beating per-question rule lookup by up to 8.98 IoU; solid, but single-run gaps and readable rules qualify the claim.
Self Improvement via Fast Tree-search
Xinghong Fu, Aravinth Kulanthaivelu, Yutaro Yamada
SIFT uses LLM-judge Bradley-Terry rankings to cut self-improvement tree-search cost; promising results, but single-run evidence and numeric inconsistencies.
A New Gap Sequence for Shellsort: RL-Driven Algorithm Discovery Beyond $N^{4/3}$
Bo Liu
RL-discovered Shellsort gap sequence narrowly beats Tokuda/Ciura on 25 large tasks; sub-4/3 proof is careful but its added gaps exist only past 10^1000.
When Hiring Becomes Agent-Mediated: Evaluating Access and Recurrence in Two-Agent R\'esum\'e Screening
Jian Gao, Hang Jiang
Claims two-agent résumé screening reallocates, not just expands, access vs one-call; confounded procedure contrast and small cohorts cap confidence.
A Unified Evaluation Framework for Trustworthy Large Language Models, Agentic AI, and Multimodal Systems
Shaina Raza, Ahmed Y. Radwan, Imran Liaquat, Kathryn Hume
A well-specified eight-dimension framework unifying LLM, agentic, and multimodal trustworthiness evaluation; conceptually strong but empirically unvalidated.
LLM-as-an-Improver: Turning Verification into Better Candidates
Akiyoshi Tomihari, Yuma Ichikawa
VRR reuses verifier feedback to generate better candidates; honest all-wrong-pool recovery shown, but gains are modest and untested for significance.
MorphIK: Morphology-Conditioned Neural Inverse Kinematics for Unknown Robots
Lennart Clasmeier, Jan Gerrit Habekost, Cornelius Weber, Stefan Wermter
Claims a flow-matching IK model that generalizes to unseen 6–9 DoF robots (~4% reach error, sub-mm after 3 DLS steps); promising but lacks external baselines.
Safety Beyond the Interface: Detecting Harm via Latent States in Large Language Models
Alizishaan Khatri, Chiquita Prabhu, Omkar Neogi
12.6M-param MLP probes on LLaMA-3.1-8B hidden states detect harmful prompts (F1 99/83/84); promising, but only vanilla prompts tested, no baselines.
An Architecture for Long-Horizon Agents: Levels, Ticks and Cascaded Intelligence
Erik Nijkamp, Anurag Koul, Egor Pakhomov, Bo Pang
A levels/ticks/cascade harness ran a 10-day autonomous RL reproduction with daily human checks; credible existence proof, but one campaign, no baselines.
QVAC Genesis III: A Large-Scale, High-Quality Open Synthetic STEM Corpus for Efficient Language Model Pre-Training
Davide Vitabile, N. Ranjan, Akshay Nambiar, Kamal K. Gupta et al.
Claims its 191B-token STEM synthetic corpus beats Cosmopedia baselines at 1.7B from-scratch scale; strong controls, but single-seed evidence tempers the gains.
The syntax and semantics of goals
David M. Abel, Mark K. Ho
A position paper mapping RL goal formalisms onto the linguistic syntax-semantics split; conceptually clarifying but adds no new theorems or evidence.
Closed-World Resolution Against Tool Hallucination in LLM Agents
Laxmipriya Ganesh Iyer
Measurement/benchmark study: closed-world resolver must precede gates to block tool hallucination; candid, but core checks are by construction and simple.
MAGS: Multi-agent Auto-formalization Guarantees Safety for Agentic Outputs
Albert Wu, Nicholas Roberts, Tzu-Heng Huang, Haoran Lin et al.
MAGS: Dafny-verified safety for 220 AI-generated programs via frozen formal semantics; candid and credible, but specs thinly audited, no baselines compared.
Working with Agentic `Teammates': When a New Organizational Actor Collides with the Human Ecosystem of Work
Rida Qadri, Remi Denton, Michael Madaio, Mahima Pushkarna et al.
In-situ study (n=17) of a proactive AI teammate in a real company: norms, relational boundaries and agency in flux; rich but small, single-firm, self-reported.
Do AI Agents Understand Computer Architecture?
Ambika Sharan, Grigory Chirkov, Soheil Abbasloo
AutoTuring isolates whether AI chip-design agents use architectural meaning; elegant design, honest limits, but 5-6 runs and one modeled GPU weaken the verdict.
Position: It is Time to Virtualize Foundation Models with a Self-evolving Operating System Layer
Suparna Bhattacharya, Tarun Kumar, Cong Xu, Satish Kumar Mopur et al.
Position paper proposing an FMOS layer to virtualize foundation models: compelling fragmentation diagnosis, but no prototype, evaluation, or evidence.