arXiv subjects

Li Wei

Publications and source records attributed to Li Wei.

At least 19 recordsLinked to original sources

arXiv2026

Building the Harness Automatically: Self-Play in Code Distills a Text Harness for Black-Box Optimization

Can an agent learn a numerical search strategy through executable practice and then transfer that strategy as text? We study low-budget black-box optimization, where unaided language models remain well below strong classical optimizers. During development, an agent repeatedly writes and evaluates optimizer programs. It then distills the resulting program and practice record once into a 197-word primary Harness A, which is frozen before evaluation. Harness A reduces Gemini Flash regret by 48\% in an independent $N=30$ study ($p<.001$), enters the GP-BO performance range on the practice family, and lowers mean regret on all three held-out BBOB landscapes. The same text improves every tested Gemini executor and transfers to Claude Sonnet, reducing regret by 43\% and 49\% ($p\leq.005$). An independent end-to-end replication produces Harness B, a different program and text at the same performance tier. The same framework also attains the lowest regret on a sealed YouTube reward-tuning production benchmark. Executable practice is thus a viable way to discover a search policy, and language a portable medium for deploying it.

cs.LG↗

arXiv2026

Multidimensional Design of Metal-Nitrogen-Carbon Electrocatalysts for Direct Propylene Epoxidation

Propylene oxide is a major industrial chemical whose production currently relies on hazardous chlorine- or peroxide-based oxidants. Direct electrochemical epoxidation using water as the oxygen source offers a sustainable alternative, but controlling oxygen-atom transfer against the competing oxygen evolution reaction remains a fundamental challenge. Here, we show that propylene epoxidation selectivity cannot be described by oxygen binding energy alone, but is jointly governed by oxygen adsorption, the potential of zero charge, and applied potential. By combining theoretical calculations with pH-field-coupled microkinetic modeling across 41 metal-nitrogen-carbon single-atom catalysts, we first identified an optimal oxygen-binding window and Co as the most favorable metal center. We then found that peripheral substituents can tune the PZC while largely preserving the optimal oxygen adsorption energetics, thereby providing an independent design dimension to further optimize the already favorable Co active site. This sequential, multidimensional design strategy identified CoPc-NH2-CNT as the optimal catalyst, delivering a record PO Faradaic efficiency of 70-80 percent for direct propylene epoxidation in aqueous electrolyte under ambient conditions. These results establish interfacial electrostatics as an independently tunable design dimension for controlling selective oxygen-atom transfer in electrocatalysis.

physics.chem-ph↗

arXiv2026

Diverge-Merge Formation and MAC Control in Structured Airspace

The rapid scaling of advanced air mobility (AAM) makes corridor-based structured airspace a promising infrastructure for high-density unmanned aerial vehicle (UAV) traffic. Formation flight can improve corridor capacity by suppressing shockwave propagation, but rigid formations become inefficient or unsafe during ramp branching, merging, and congestion. To address this problem, this paper proposes a task-driven diverge-merge control framework for UAV formations in structured airspace. At the beginning, a corridor-ramp branching structured airspace model is established to characterize the traffic dynamics and spatial constraints. Building upon this, a fast task-driven clustering mechanism integrates spatial connectivity, flight intent, and aerial task interactions to enable real-time diverge and merge for ramp branching and traffic reshaping. To make the diverge-merge reconfigurations executable at the media access control (MAC) layer of the formation, a cluster-aware distributed time division multiple access (CAD-TDMA) protocol is further designed. It protects intra-cluster control synchronization while conservatively reusing low-risk inter-cluster slots. Simulation results show that the proposed diverge-merge algorithm maintains near-zero geometrical misclassification under severe physical overlapping and congestion. With the formation diverge-merge traces, CAD-TDMA achieves the best delay--loss--throughput tradeoff over fixed TDMA and WiFi MAC. It shows that the proposed formation control framework can jointly support real-time formation reconfiguration and reliable communication in corridor-ramp structured airspace.

cs.NI↗

arXiv2026

TokenMinds: Pretrained User Tokens and Embeddings for User Understanding in Large Recommender Systems

User modeling in industrial recommender systems typically produces dense embeddings, which suffer from representational constraints inherent to fixed-dimensional vectors. An emerging alternative for discrete user representation -- using LLMs to generate text-based user tokens -- captures topical co-occurrences rather than deep sequential behavior dynamics and produces outputs that are difficult to ground to item attributes. Meanwhile, Semantic ID (SID) based item tokenization has proven effective for improving generalization in generative recommendation, yet discrete SID-based representations for users remain largely unexplored. We propose TokenMinds, an industrial-scale system that extends the PLUM framework from item retrieval to user modeling, generating both discrete SID-based user tokens and dense user embeddings via an encoder-decoder architecture adapted from pre-trained LLMs. This dual-output design provides the complementary benefits of discrete, semantically grounded user representations while maintaining compatibility with existing downstream models that rely on dense embeddings. Additionally, the shared SID vocabulary naturally extends to cross-scenario modeling: by unifying long-form and short-form video behaviors into a single model, we substantially reduce training and serving costs. We validate TokenMinds through extensive offline experiments and live launches on multiple YouTube surfaces, served on full user traffic (billions of users) via an asynchronous infrastructure that decouples representation generation from downstream scoring. Focusing on ranking as the primary downstream use case, our results confirm the practical viability of SID-based user tokens at industrial scale and demonstrate that tokens and dense embeddings provide complementary value across different production ranking systems.

cs.IR↗

arXiv2026

TelecomGPT-R1: A Unified Open-Source Reasoner for the Telecom Stack

Telecommunications is a high-leverage domain for large language model (LLM)-based reasoning because routine engineering workflows require joint grounding in normative specifications, operational telemetry, vendor-specific fault evidence, and exact RF/network calculations. However, current LLM integration in telecom remains bottlenecked by a two-sided capability gap: generic reasoners often lack telecom-specific grounding, while domain-specific telecom LLMs remain limited in structured, multi-step reasoning. To bridge this gap, we release TelecomGPT-R1-9B, a unified open-source telecom reasoner that ranks top-performing on the GSMA open telco leaderboard. Specifically, we curate a 67,427-example supervised fine-tuning (SFT) corpus organized around four complementary reasoning axes: protocol, knowledge, modeling, and fault. The corpus is built from axis-matched public web sources and enhanced through axis-specific chain-of-thought (CoT) generation and prefix-continuation self-validation. Starting from Qwen3.5-9B, we further develop a two-stage post-training recipe. First, multi-teacher low-rank adaptation (LoRA)-based SFT injects telecom knowledge and induces axis-specific reasoning formats. Second, group relative policy optimization (GRPO), stabilized by decoupled clip and dynamic sampling policy optimization (DAPO), optimizes the policy using four axis-aligned binary verifier rewards. Across seven public telecom benchmarks, TelecomGPT-R1-9B ranks first among open-source telecom LLMs and achieves a seven-axis mean comparable to state-of-the-art closed-source frontier reasoners.

cs.CL↗

arXiv2026

Token Factory: Efficiently Integrating Diverse Signals into Large Recommendation Models

Large Recommendation Models (LRMs) have demonstrated promising capabilities in industry-scale recommendation tasks. However, holistically integrating traditional signals into these transformer-based architectures effectively and efficiently remains a major challenge. Conventional approaches that "textualize" these signals directly or create discrete item representations often lead to excessively long prompts, substantial memory footprints, and high computational overhead. To overcome these limitations, we propose "Token Factory", a framework designed to transform traditional signals into "soft tokens" that can be directly processed by LRMs. This approach enables efficient integration and compression of heterogeneous input features, preventing prompt length explosion while enhancing model performance. We detail the architecture of Token Factory and present experimental results validating its effectiveness in a production-scale recommendation environment.

cs.IR↗

arXiv2026

ORBIT: Preserving Foundational Language Capabilities in GenRetrieval via Origin-Regulated Merging

Despite the rapid advancements in large language model (LLM) development, fine-tuning them for specific tasks often results in the catastrophic forgetting of their general, language-based reasoning abilities. This work investigates and addresses this challenge in the context of the Generative Retrieval (GenRetrieval) task. During GenRetrieval fine-tuning, we find this forgetting occurs rapidly and correlates with the distance between the fine-tuned and original model parameters. Given these observations, we propose ORBIT, a novel approach that actively tracks the distance between fine-tuned and initial model weights, and uses a weight averaging strategy to constrain model drift during GenRetrieval fine-tuning when this inter-model distance exceeds a maximum threshold. Our results show that ORBIT retains substantial text and retrieval performance by outperforming both common continual learning baselines and related regularization methods that also employ weight averaging.

cs.CL↗

arXiv2026

Spin State versus Potential of Zero Charge as Predictors of Density-Dependent Oxygen Reduction in M-N-C Electrocatalysts

Metal-site density strongly influences oxygen reduction activity and selectivity in M-N-C electrocatalysts, but the descriptors that predict these trends remain under debate. Here, we compare spin state and the potential of zero charge as predictors of density-dependent oxygen reduction behavior in Fe-N-C and Co-N-C catalysts. Using constrained-magnetization calculations combined with Landau analysis, we find that the ground-state magnetic moments vary only weakly across a broad range of metal-site densities, suggesting that magnetic descriptors alone cannot account for the pronounced performance changes. In contrast, explicit-solvent simulations reveal systematic density-dependent shifts in PZC, which alter the interfacial electric field and thereby modulate field-sensitive adsorption energetics of ORR intermediates. Incorporating these PZC shifts into a pH-field-coupled microkinetic model captures the density-dependent activity trends and reproduces the experimentally observed increase in two-electron selectivity at lower site densities under acidic conditions. Experimental PZC measurements further support the predicted trend. Together, these results show that PZC is a more effective predictor than spin state for density-dependent oxygen reduction activity and selectivity in M-N-C electrocatalysts.

cond-mat.mtrl-sci↗

arXiv2026

Zero-shot Cross-domain Knowledge Distillation: A Case study on YouTube Music

Knowledge Distillation (KD) has been widely used to improve the quality of latency sensitive models serving live traffic. However, applying KD in production recommender systems with low traffic is challenging: the limited amount of data restricts the teacher model size, and the cost of training a large dedicated teacher may not be justified. Cross-domain KD offers a cost-effective alternative by leveraging a teacher from a data-rich source domain, but introduces unique technical difficulties, as the features, user interfaces, and prediction tasks can significantly differ. We present a case study of using zero-shot cross-domain KD for multi-task ranking models, transferring knowledge from a (100x) large-scale video recommendation platform (YouTube) to a music recommendation application with significantly lower traffic. We share offline and live experiment results and present findings evaluating different KD techniques in this setting across two ranking models on the music app. Our results demonstrate that zero-shot cross-domain KD is a practical and effective approach to improve the performance of ranking models on low traffic surfaces.

cs.IR↗

arXiv2026

Self-Evolving Recommendation System: End-To-End Autonomous Model Optimization With LLM Agents

Optimizing large-scale machine learning systems, such as recommendation models for global video platforms, requires navigating a massive hyperparameter search space and, more critically, designing sophisticated optimizers, architectures, and reward functions to capture nuanced user behaviors. Achieving substantial improvements in these areas is a non-trivial task, traditionally relying on extensive manual iterations to test new hypotheses. We propose a self-evolving system that leverages Large Language Models (LLMs), specifically those from Google's Gemini family, to autonomously generate, train, and deploy high-performing, complex model changes within an end-to-end automated workflow. The self-evolving system consists of an Offline Agent (Fast Loop) that performs high-throughput hypothesis generation to optimize for proxy metrics, and an Online Agent (Slow Loop) that validates candidates against delayed north star business metrics in live production. Our agents act as specialized Machine Learning Engineers (MLEs): they exhibit deep reasoning capabilities, discovering novel improvements in optimization algorithms and model architecture, and formulating innovative reward functions that target long-term user engagement. The effectiveness of this approach is demonstrated through several successful production launches at YouTube, confirming that autonomous, LLM-driven evolution can surpass traditional engineering workflows in both development velocity and model performance.

cs.LG↗

arXiv2026

Learning to Alleviate Familiarity Bias in Video Recommendation

Modern video recommendation systems aim to optimize user engagement and platform objectives, yet often face structural exposure imbalances caused by behavioral biases. In this work, we focus on the post-ranking stage and present LAFB (Learning to Alleviate Familiarity Bias), a lightweight and model-agnostic framework designed to mitigate familiarity bias in recommendation outputs. LAFB models user-content familiarity using discrete and continuous interaction features, and estimates personalized debiasing factors to adjust user rating prediction scores, thereby reducing the dominance of familiar content in the final ranking. We conduct large-scale offline evaluations and online A/B testing in a real-world recommendation system, under a unified serving stack that also compares LAFB with deployable popularity-oriented remedies. Results show that LAFB increases novel watch-time share and improves exposure for emerging creators and overall content diversity, while maintaining stable overall watch time and short-term satisfaction. LAFB has already been launched in the post-ranking stage of YouTube's recommendation system, demonstrating its effectiveness in real-world applications.

cs.IR↗

arXiv2026

LongCat-Flash-Thinking-2601 Technical Report

We introduce LongCat-Flash-Thinking-2601, a 560-billion-parameter open-source Mixture-of-Experts (MoE) reasoning model with superior agentic reasoning capability. LongCat-Flash-Thinking-2601 achieves state-of-the-art performance among open-source models on a wide range of agentic benchmarks, including agentic search, agentic tool use, and tool-integrated reasoning. Beyond benchmark performance, the model demonstrates strong generalization to complex tool interactions and robust behavior under noisy real-world environments. Its advanced capability stems from a unified training framework that combines domain-parallel expert training with subsequent fusion, together with an end-to-end co-design of data construction, environments, algorithms, and infrastructure spanning from pre-training to post-training. In particular, the model's strong generalization capability in complex tool-use are driven by our in-depth exploration of environment scaling and principled task construction. To optimize long-tailed, skewed generation and multi-turn agentic interactions, and to enable stable training across over 10,000 environments spanning more than 20 domains, we systematically extend our asynchronous reinforcement learning framework, DORA, for stable and efficient large-scale multi-environment training. Furthermore, recognizing that real-world tasks are inherently noisy, we conduct a systematic analysis and decomposition of real-world noise patterns, and design targeted training procedures to explicitly incorporate such imperfections into the training process, resulting in improved robustness for real-world applications. To further enhance performance on complex reasoning tasks, we introduce a Heavy Thinking mode that enables effective test-time scaling by jointly expanding reasoning depth and width through intensive parallel thinking.

cs.AI↗

arXiv2025

Efficient Context Scaling with LongCat ZigZag Attention

We introduce LongCat ZigZag Attention (LoZA), which is a sparse attention scheme designed to transform any existing full-attention models into sparse versions with rather limited compute budget. In long-context scenarios, LoZA can achieve significant speed-ups both for prefill-intensive (e.g., retrieval-augmented generation) and decode-intensive (e.g., tool-integrated reasoning) cases. Specifically, by applying LoZA to LongCat-Flash during mid-training, we serve LongCat-Flash-Exp as a long-context foundation model that can swiftly process up to 1 million tokens, enabling efficient long-term reasoning and long-horizon agentic capabilities.

cs.CL↗

arXiv2025

Selecting User Histories to Generate LLM Users for Cold-Start Item Recommendation

Large Language Models (LLMs) have demonstrated remarkable capabilities in reasoning, generalization, and simulating human-like behavior across a wide range of tasks. These strengths present new opportunities to enhance traditional recommendation systems (RS), especially in the cold-start item scenario where newly introduced items lack interactions. Existing works have used LLMs to address cold-start issues in traditional RS through data augmentation, but they have limitations. One recent work directly addresses this issue by prompting LLMs to generate augmented interaction data between randomly sampled users and cold-start items. Then, they train the traditional RS with augmented data, incorporating collaborative signals for cold-start items. Although they use LLMs to provide cold-start items with feedback, they use partial user histories, which does not allow the LLM to fully emulate the user. Furthermore, randomly selecting users is not optimal for augmentation. To address these challenges, we leverage the LLM as a user and develop a reinforcement learning (RL) framework that trains a policy to select users for augmentation, optimizing for cold-start item performance after augmented training. The policy model learns to select users for cold-start item data augmentation based on their behavioral features and histories. To optimize user selection for cold-start item performance, we employ a policy gradient method that updates the policy in the direction of actions that lead to high rewards. Experiments on Amazon Product Review datasets show substantial gains in cold-start item recall, demonstrating the effectiveness of our method as a scalable, serving-efficient augmentation strategy for modern RS.

cs.IR↗

arXiv2025

WirelessMathLM: Teaching Mathematical Reasoning for LLMs in Wireless Communications with Reinforcement Learning

Large language models (LLMs) excel at general mathematical reasoning but fail catastrophically on specialized technical mathematics. In wireless communications, where problems require precise manipulation of information-theoretic bounds, optimization constraints, and signal processing formulations, even state-of-the-art models struggle to achieve competent performance. We present WirelessMathLM, demonstrating that compact models (0.5B-7B parameters) can match or exceed much larger models through domain-specific reinforcement learning with verifiable rewards. Our key insight is that wireless mathematics problems possess a unique property--verifiable correctness--that enables effective reinforcement learning without human feedback. We construct WirelessMathBench-XL, a comprehensive benchmark of 4,027 problems from 970 papers. Using Group Relative Policy Optimization (GRPO) with binary verification rewards, we train models directly from base checkpoints without supervised warm-start. Our 7B model achieves 39.5% accuracy on WirelessMathBench-XL, approaching GPT-4o (40.4%) while using about 100 times fewer parameters than DeepSeek-R1 (671B, 57.4%). Remarkably, GRPO training nearly doubles performance across all model scales (0.5B +11%, 3B +103%, 7B +81%), with positive transfer to general mathematics benchmarks--our models gain +8.4 points on average across MATH, Minerva-Math, OlympiadBench, AMC, and AIME without any training on these tasks.

cs.LG↗

arXiv2025

Introducing LongCat-Flash-Thinking: A Technical Report

We present LongCat-Flash-Thinking, an efficient 560-billion-parameter open-source Mixture-of-Experts (MoE) reasoning model. Its advanced capabilities are cultivated through a meticulously crafted training process, beginning with long Chain-of-Thought (CoT) data cold-start and culminating in large-scale Reinforcement Learning (RL). We first employ a well-designed cold-start training strategy, which significantly enhances the reasoning potential and equips the model with specialized skills in both formal and agentic reasoning. Then, a core innovation is our domain-parallel training scheme, which decouples optimization across distinct domains (e.g., STEM, Code, Agentic) and subsequently fuses the resulting expert models into a single, nearly Pareto-optimal model. This entire process is powered by our Dynamic ORchestration for Asynchronous rollout (DORA) system, a large-scale RL framework that delivers a greater than threefold training speedup over synchronous methods on tens of thousands of accelerators. As a result, LongCat-Flash-Thinking achieves state-of-the-art performance among open-source models on a suite of complex reasoning tasks. The model exhibits exceptional efficiency in agentic reasoning, reducing average token consumption by 64.5% (from 19, 653 to 6, 965) on AIME-25, without degrading task accuracy. We release LongCat-Flash-Thinking to promote further advances in reasoning systems and agentic AI research.

cs.AI↗

arXiv2025

Deploying UDM Series in Real-Life Stuttered Speech Applications: A Clinical Evaluation Framework

Stuttered and dysfluent speech detection systems have traditionally suffered from the trade-off between accuracy and clinical interpretability. While end-to-end deep learning models achieve high performance, their black-box nature limits clinical adoption. This paper looks at the Unconstrained Dysfluency Modeling (UDM) series-the current state-of-the-art framework developed by Berkeley that combines modular architecture, explicit phoneme alignment, and interpretable outputs for real-world clinical deployment. Through extensive experiments involving patients and certified speech-language pathologists (SLPs), we demonstrate that UDM achieves state-of-the-art performance (F1: 0.89+-0.04) while providing clinically meaningful interpretability scores (4.2/5.0). Our deployment study shows 87% clinician acceptance rate and 34% reduction in diagnostic time. The results provide strong evidence that UDM represents a practical pathway toward AI-assisted speech therapy in clinical environments.

cs.SD↗

arXiv2025

ACT: Automated Constraint Targeting for Multi-Objective Recommender Systems

Recommender systems often must maximize a primary objective while ensuring secondary ones satisfy minimum thresholds, or "guardrails." This is critical for maintaining a consistent user experience and platform ecosystem, but enforcing these guardrails despite orthogonal system changes is challenging and often requires manual hyperparameter tuning. We introduce the Automated Constraint Targeting (ACT) framework, which automatically finds the minimal set of hyperparameter changes needed to satisfy these guardrails. ACT uses an offline pairwise evaluation on unbiased data to find solutions and continuously retrains to adapt to system and user behavior changes. We empirically demonstrate its efficacy and describe its deployment in a large-scale production environment.

cs.IR↗