arXiv subjects

Wanyi Chen

Publications and source records attributed to Wanyi Chen.

15 recordsLinked to original sources

arXiv2026

MERA: Model Evolution and Routing with Skill Adaptation for Agentic Systems at Scale

LLM agents execute heterogeneous sequences of model calls within a single task: some invocations require careful reasoning, while others are structured steps such as formatting or tool-argument construction. Prior routing methods exploit this asymmetry by assigning easy invocations to a cheaper small model and difficult ones to a large model. Such policies reduce inference cost, but they leave the small model's capability unchanged, so attainable savings remain bounded by the work the student can already solve. MERA instead improves the small model itself, using a single model invocation as the unit of adaptation. In each cycle, MERA replays failed student invocations to obtain execution-verified teacher demonstrations, distills recurring procedures into an iteratively updated SkillBook, and fine-tunes a student LoRA adapter via supervised learning and optional GRPO. Routing serves as supporting machinery for deployment: the improved student is served behind a cost-calibrated router with verifier-backed fallback, and a candidate SkillBook, adapter, or router is admitted only when joint replay preserves task quality. Empirically, four-cycle adaptation raises Qwen2.5-Coder-1.5B from 28.7% to 49.7% pass on held-out HumanEval+MBPP. Under verifier-backed fallback, the deployed policy retains 88.3% pass at 60.8% of always-Luna cost. On TAU-2, a fine-tuned Qwen3.5-2B improves from 14/35 to 18/35 and matches an unadapted 4B model. These results indicate that verifier-backed multi-cycle adaptation can increase small-model capability, rather than only routing around a fixed student.

cs.LG↗

arXiv2026

Attributing Structured-Output Gains in Function Calling: Interface Alignment versus Procedural Transfer

Structured-output benchmarks reward both task decisions and interface compliance, so prompt-induced function-calling gains require attribution before they can be interpreted as transferable skill. We introduce a four-layer gain-attribution protocol for prompt-prepended skill injection, combining canonicalized rescoring, format-only controls, repaired/balanced induction, and portability checks. Applied to the Berkeley Function Calling Leaderboard (BFCL) and scoped with API-Bank, MATH-500, and MultiHop-RAG, the protocol shows that several apparent gains are better attributed to interface alignment than to procedural transfer: format-only prompts match or exceed full skills in key BFCL cells, repaired/balanced induction removes the largest sub-frontier gains, and API-Bank target-native gains are matched within 0.5 percentage points (pp) by length-matched generic procedural prompts. These findings treat format compliance as a useful engineering capability while clarifying what a structured-output score certifies. We release BFCL-CANONICAL and recommend canonicalized metrics, balanced induction, and format-only baselines for function-calling skill-gain attribution. Code and data are available at https://github.com/couragec/skill-injection-attribution.

cs.SE↗

arXiv2026

TwinRouterBench: Fast Static and Live Dynamic Evaluation for Realistic Agentic LLM Routing

LLM routing matters most in long-horizon applications such as coding agents, deep research systems, and computer-use agents, where a single user request triggers many model calls. Routing each call to the cheapest sufficient model can cut costs without sacrificing quality, yet existing router benchmarks evaluate routers only on one-shot prompts. They never expose the router-visible prefix at an intermediate agent step, never test whether a cheaper replacement preserves downstream task success, and often rely on online LLM judges at evaluation time. We introduce TwinRouterBench, a step-level routing benchmark with two tracks. The static track provides 970 router-visible prefixes from 520 instances across SWE-bench, BFCL, mtRAG, QMSum, and PinchBench, each paired with an execution-verified target tier estimated under a released downgrade-and-cascade protocol; scoring is deterministic arithmetic over tier labels, trajectory membership, and token costs, with no online evaluator-side LLM judge. The dynamic track supplies a harness that runs routers on the full 500-case SWE-bench Verified suite; in this paper we report a 100-case held-out evaluation disjoint from the static SWE supervision split. At each LLM call the router selects a concrete model from a locked pool, and success is measured by official task resolution and realized API spend. The two tracks support fast offline iteration followed by end-to-end validation under live agent execution. Code and data are available at https://github.com/CommonstackAI/TwinRouterBench.

cs.LG↗

arXiv2026

Agent^2 RL-Bench: Can LLM Agents Engineer Agentic RL Post-Training?

We introduce Agent2 RL-Bench, a compact diagnostic benchmark for evaluating agentic RL post-training, which tests whether LLM agents can autonomously design, implement, debug, and execute post-training pipelines that improve foundation models. RL post-training increasingly drives model alignment and specialization, yet existing benchmarks are largely static, rewarding supervised fine-tuning or script generation without assessing an agent's ability to close an interactive RL loop. Agent2 RL-Bench provides a unified agent-facing interface: each run starts from an isolated workspace containing a base model, task data, instructions, and a grading API, and agents must iterate within a fixed budget by training models and submitting artifacts for evaluation. The benchmark spans six tasks across three levels, from static rule-based training to judge-based optimization and closed-loop online RL with trajectory collection. Two diagnostic skills, namely runtime recording and post-hoc summarization, enable structured analysis of agent behavior, facilitating smooth and effective iteration of the benchmark's evaluation framework. Across five agent systems and six driver LLMs, agents show intelligent behavior but clear limitations: one RL-oriented run improves ALFWorld from 4.85 to 93.28 via SFT warm-up and GRPO with online rollouts, yet DeepSearchQA remains difficult, most successful routes rely on supervised pipelines, and interactive outcomes show large single-run differences across agent stacks. Overall, Agent2 RL-Bench shows that current agents can sometimes engineer online RL, but stable agent-driven RL post-training remains rare under fixed budgets. It also demonstrates that our benchmark provides a strong and effective evaluation framework for future research in this direction. Code is available at https://github.com/microsoft/RD-Agent/blob/main/rdagent/scenarios/rl/autorl_bench/README.md

cs.AI↗

arXiv2026

QHap: Quantum-Inspired Haplotype Phasing

Haplotype phasing, the process of resolving parental allele inheritance patterns in diploid genomes, is critical for precision medicine and population genetics, yet the underlying optimization is NP-hard, posing a scalability challenge. To address this, we introduce QHap, a haplotype phasing algorithm that leverages quantum-annealing-inspired optimization. By reformulating haplotype phasing as a Max-Cut problem and deploying a GPU-accelerated ballistic simulated bifurcation solver, QHap accelerates phasing while maintaining accuracy comparable to established phasing tools. On the highly polymorphic human major histocompatibility complex region, QHap demonstrates 4- to 20-fold acceleration over HapCUT2 and WhatsHap with zero switch error across multiple long-read sequencing platforms. The framework implements two strategies: a read-based method for regional phasing, and a single nucleotide polymorphism-based method that, through quality-weighted probabilistic edge construction, efficiently scales to chromosome-scale tasks. Integration of Pore-C chromatin conformation capture data increases the haplotype N50 by up to 15-fold, enabling near-chromosome-scale haplotype reconstruction. QHap demonstrates that quantum-inspired algorithms operating on classical hardware offer a promising approach to addressing the growing computational demands of sequencing data, establishing a new paradigm for applying physics-inspired optimization to fundamental challenges in computational genomics.

q-bio.GN↗

arXiv2026

AOI: Turning Failed Trajectories into Training Signals for Autonomous Cloud Diagnosis

Large language model (LLM) agents offer a promising data-driven approach to automating Site Reliability Engineering (SRE), yet their enterprise deployment is constrained by three challenges: restricted access to proprietary data, unsafe action execution under permission-governed environments, and the inability of closed systems to improve from failures. We present AOI (Autonomous Operations Intelligence), a trainable multi-agent framework formulating automated operations as a structured trajectory learning problem under security constraints. Our approach integrates three key components. First, a trainable diagnostic system applies Group Relative Policy Optimization (GRPO) to distill expert-level knowledge into locally deployed open-source models, enabling preference-based learning without exposing sensitive data. Second, a read-write separated execution architecture decomposes operational trajectories into observation, reasoning, and action phases, allowing safe learning while preventing unauthorized state mutation. Third, a Failure Trajectory Closed-Loop Evolver mines unsuccessful trajectories and converts them into corrective supervision signals, enabling continual data augmentation. Evaluated on the AIOpsLab benchmark, our contributions yield cumulative gains. (1) The AOI runtime alone achieves 66.3% best@5 success on all 86 tasks, outperforming the prior state-of-the-art (41.9%) by 24.4 points. (2) Adding Observer GRPO training, a locally deployed 14B model reaches 42.9% avg@1 on 63 held-out tasks with unseen fault types, surpassing Claude Sonnet 4.5. (3) The Evolver converts 37 failed trajectories into diagnostic guidance, improving end-to-end avg@5 by 4.8 points while reducing variance by 35%.

cs.LG↗

arXiv2026

ECHO-2: A Large-Scale Distributed Rollout Framework for Cost-Efficient Reinforcement Learning

Reinforcement learning (RL) is a critical stage in post-training large language models (LLMs), involving repeated interaction between rollout generation, reward evaluation, and centralized learning. Distributing rollout execution offers opportunities to leverage more cost-efficient inference resources, but introduces challenges in wide-area coordination and policy dissemination. We present ECHO-2, a distributed RL framework for post-training with remote inference workers and non-negligible dissemination latency. ECHO-2 combines centralized learning with distributed rollouts and treats bounded policy staleness as a user-controlled parameter, enabling rollout generation, dissemination, and training to overlap. We introduce an overlap-based capacity model that relates training time, dissemination latency, and rollout throughput, yielding a practical provisioning rule for sustaining learner utilization. To mitigate dissemination bottlenecks and lower cost, ECHO-2 employs peer-assisted pipelined broadcast and cost-aware activation of heterogeneous workers. Experiments on GRPO post-training of LLMs ranging from 4B to 32B parameters under real wide-area bandwidth regimes show that ECHO-2 significantly improves cost efficiency while preserving RL reward comparable to strong baselines.

cs.LG↗

arXiv2026

EVM-QuestBench: An Execution-Grounded Benchmark for Natural-Language Transaction Code Generation

Large language models are increasingly applied to various development scenarios. However, in on-chain transaction scenarios, even a minor error can cause irreversible loss for users. Existing evaluations often overlook execution accuracy and safety. We introduce EVM-QuestBench, an execution-grounded benchmark for natural-language transaction-script generation on EVM-compatible chains. The benchmark employs dynamic evaluation: instructions are sampled from template pools, numeric parameters are drawn from predefined intervals, and validators verify outcomes against these instantiated values. EVM-QuestBench contains 107 tasks (62 atomic, 45 composite). Its modular architecture enables rapid task development. The runner executes scripts on a forked EVM chain with snapshot isolation; composite tasks apply step-efficiency decay. We evaluate 20 models and find large performance gaps, with split scores revealing persistent asymmetry between single-action precision and multi-step workflow completion. Code: https://anonymous.4open.science/r/bsc_quest_bench-A9CF/.

cs.CL↗

arXiv2025

Speculative Decoding in Decentralized LLM Inference: Turning Communication Latency into Computation Throughput

Speculative decoding accelerates large language model (LLM) inference by using a lightweight draft model to propose tokens that are later verified by a stronger target model. While effective in centralized systems, its behavior in decentralized settings, where network latency often dominates compute, remains under-characterized. We present Decentralized Speculative Decoding (DSD), a plug-and-play framework for decentralized inference that turns communication delay into useful computation by verifying multiple candidate tokens in parallel across distributed nodes. We further introduce an adaptive speculative verification strategy that adjusts acceptance thresholds by token-level semantic importance, delivering an additional 15% to 20% end-to-end speedup without retraining. In theory, DSD reduces cross-node communication cost by approximately (N-1)t1(k-1)/k, where t1 is per-link latency and k is the average number of tokens accepted per round. In practice, DSD achieves up to 2.56x speedup on HumanEval and 2.59x on GSM8K, surpassing the Eagle3 baseline while preserving accuracy. These results show that adapting speculative decoding for decentralized execution provides a system-level optimization that converts network stalls into throughput, enabling faster distributed LLM inference with no model retraining or architectural changes.

cs.DC↗

arXiv2025

From Perception to Cognition: A Survey of Vision-Language Interactive Reasoning in Multimodal Large Language Models

Multimodal Large Language Models (MLLMs) strive to achieve a profound, human-like understanding of and interaction with the physical world, but often exhibit a shallow and incoherent integration when acquiring information (Perception) and conducting reasoning (Cognition). This disconnect leads to a spectrum of reasoning failures, with hallucination being the most prominent. Collectively, these issues expose a fundamental challenge: the ability to process pixels does not yet confer the ability to construct a coherent, credible internal world model. To systematically dissect and address this challenge, this survey introduces a novel and unified analytical framework: ``From Perception to Cognition." We deconstruct the complex process of vision-language interactive understanding into two interdependent layers: Perception, the foundational ability to accurately extract visual information and achieve fine-grained alignment with textual instructions; and Cognition, the higher-order capability for proactive, multi-step, goal-oriented reasoning built upon this perceptual foundation, the core of which is the formation of a dynamic observe-think-verify reasoning loop. Guided by this framework, this paper systematically analyzes the key bottlenecks of current MLLMs at both layers. It surveys the landscape of cutting-edge methods designed to address these challenges, spanning from techniques that enhance low-level visual representations to those that improve high-level reasoning paradigms. Furthermore, we review critical benchmarks and delineate future research directions. This survey aims to provide the research community with a clear, structured perspective for understanding the intrinsic limitations of current MLLMs and to illuminate the path toward building next-generation models capable of deep reasoning and a genuine understanding of the world.

cs.AI↗

arXiv2025

Assessing LLM code generation quality through path planning tasks

As LLM-generated code grows in popularity, more evaluation is needed to assess the risks of using such tools, especially for safety-critical applications such as path planning. Existing coding benchmarks are insufficient as they do not reflect the context and complexity of safety-critical applications. To this end, we assessed six LLMs' abilities to generate the code for three different path-planning algorithms and tested them on three maps of various difficulties. Our results suggest that LLM-generated code presents serious hazards for path planning applications and should not be applied in safety-critical contexts without rigorous testing.

cs.SE↗

arXiv2025

To impute or not to impute: How machine learning modelers treat missing data

Missing data is prevalent in tabular machine learning (ML) models, and different missing data treatment methods can significantly affect ML model training results. However, little is known about how ML researchers and engineers choose missing data treatment methods and what factors affect their choices. To this end, we conducted a survey of 70 ML researchers and engineers. Our results revealed that most participants were not making informed decisions regarding missing data treatment, which could significantly affect the validity of the ML models trained by these researchers. We advocate for better education on missing data, more standardized missing data reporting, and better missing data analysis tools.

cs.LG↗

arXiv2024

Can LLMs plan paths in the real world?

As large language models (LLMs) increasingly integrate into vehicle navigation systems, understanding their path-planning capability is crucial. We tested three LLMs through six real-world path-planning scenarios in various settings and with various difficulties. Our experiments showed that all LLMs made numerous errors in all scenarios, revealing that they are unreliable path planners. We suggest that future work focus on implementing mechanisms for reality checks, enhancing model transparency, and developing smaller models.

cs.AI↗

arXiv2024

A Safe Screening Rule with Bi-level Optimization of $ν$ Support Vector Machine

Support vector machine (SVM) has achieved many successes in machine learning, especially for a small sample problem. As a famous extension of the traditional SVM, the $ν$ support vector machine ($ν$-SVM) has shown outstanding performance due to its great model interpretability. However, it still faces challenges in training overhead for large-scale problems. To address this issue, we propose a safe screening rule with bi-level optimization for $ν$-SVM (SRBO-$ν$-SVM) which can screen out inactive samples before training and reduce the computational cost without sacrificing the prediction accuracy. Our SRBO-$ν$-SVM is strictly deduced by integrating the Karush-Kuhn-Tucker (KKT) conditions, the variational inequalities of convex problems and the $ν$-property. Furthermore, we develop an efficient dual coordinate descent method (DCDM) to further improve computational speed. Finally, a unified framework for SRBO is proposed to accelerate many SVM-type models, and it is successfully applied to one-class SVM. Experimental results on 6 artificial data sets and 30 benchmark data sets have verified the effectiveness and safety of our proposed methods in supervised and unsupervised tasks.

cs.LG↗

arXiv2024

Subjectivity in Unsupervised Machine Learning Model Selection

Model selection is a necessary step in unsupervised machine learning. Despite numerous criteria and metrics, model selection remains subjective. A high degree of subjectivity may lead to questions about repeatability and reproducibility of various machine learning studies and doubts about the robustness of models deployed in the real world. Yet, the impact of modelers' preferences on model selection outcomes remains largely unexplored. This study uses the Hidden Markov Model as an example to investigate the subjectivity involved in model selection. We asked 33 participants and three Large Language Models (LLMs) to make model selections in three scenarios. Results revealed variability and inconsistencies in both the participants' and the LLMs' choices, especially when different criteria and metrics disagree. Sources of subjectivity include varying opinions on the importance of different criteria and metrics, differing views on how parsimonious a model should be, and how the size of a dataset should influence model selection. The results underscore the importance of developing a more standardized way to document subjective choices made in model selection processes.

cs.LG↗