Searcharxiv⌕ Search

arXiv subjects

Mo Li

Publications and source records attributed to Mo Li.

At least 19 recordsLinked to original sources
arXiv2026

Data-Adaptive Multi-Bin Rarefying for Alpha Diversity Association Analysis under Heterogeneous Library Sizes

Alpha diversity association analysis is sensitive to heterogeneous library sizes, and overall rarefying to a common depth can either discard many reads from high-depth samples or exclude low-depth samples. We propose MBRarefy, a data-adaptive framework for selecting multi-bin rarefying designs. The method chooses library size cutpoints by minimizing an objective that combines residual within-bin library size dependence with a fragmentation penalty, subject to a minimum bin size constraint. The framework supports both fixed-K and varying-K cutpoint selection. In T-cell receptor (TCR) motivated simulations, the primary fixed-K procedure controlled depth-induced false positives and improved power relative to overall rarefying when group and depth confounding was absent. Applications to TCR immunosequencing and wild baboon gut microbiome data illustrate its ability to reduce residual library size dependence while improving read and sample retention.

stat.CO↗
arXiv2026

GRP v0.1 Technical Report

Industrial recommendation systems rely on multi-stage cascades whose retrieval, ranking, and serving components are difficult to replace jointly. We present GRP, a generative recommendation framework that combines retrieval, ranking, and reward modeling in a single encoder-decoder model, and evaluate a progressive path toward end-to-end recommendation. The model generates multimodal Semantic IDs and scores candidates with a jointly trained ranking module. The frozen ranking module then supplies rewards for reinforcement-learning post-training. We introduce mGRPO, which adds a reference-anchored margin to reward optimization to preserve the likelihood of logged targets. Offline experiments examine history encoding, model capacity allocation, event selection, tokenization, and reward discrimination. Serving optimizations reduce end-to-end retrieval latency by 69%. Online experiments evaluate the model as a retrieval source, with early-ranking bypass, and with replacement of weaker sources. In a retrieval-only comparison, view time increases by 0.46% and shares by 0.77% relative to production. A separate comparison combining bypass and source replacement yields increases of 0.82% in view time and 2.56% in shares, with neutral platform-level guardrails. These results support progressive deployment while identifying remaining gaps in ranking quality and performance across recommendation metrics.

cs.IR↗
arXiv2026

Skill Training with Corruption and Reconstruction Loop

Large Language Models (LLMs) often struggle in highly specialized domains. Rather than parameter-level adaptation of LLMs which is costly and difficult to interpret, external skills (often defined as text files) have been recently proposed to augment LLMs for specialized domains. However, such skills rely on costly active human annotations or passive summarization of high-quality examples. In this paper, we propose a self-supervised approach for agent self-evolution that learns domain-specific skills directly from existing high-quality human artifacts, without additional human annotations or external rewards. Inspired by diffusion models, our approach follows a forward-loss-backward process to reconstruct human artifacts by iteratively learning the agent's external skill library rather than updating its model parameters. Experiments on short-drama screenwriting demonstrate that our approach enables agents to autonomously extract generalizable writing skills from human-authored scripts and substantially improve domain-specific generation quality. Our approach provides a scalable paradigm for agents to continuously learn many kinds of complex skills from existing high-quality human artifacts. Code and project page: https://github.com/skilltraining-project/skill-training and https://skilltraining-project.github.io

cs.CL↗
arXiv2026

LLM-Based Generative Retrieval for Snapchat Content Recommendation

Pretrained large language models (LLMs) are promising retrieval engines because they combine rich semantic priors, strong sequence modeling capabilities, and favorable scaling behavior. However, turning a pretrained LLM into a generative retriever in production deployment raises several challenges: the model must learn an internal item vocabulary that was absent from pretraining, and generate valid item identifiers under strict latency and cost constraints. We address these challenges through the design and launch of SnapLGR, an LLM-based generative retrieval system for short-video recommendation at Snapchat. The system is built around three main designs. First, we construct semantic identifiers (SIDs) from multimodal item embeddings and enhance them with Personalized PageRank (PPR)-based co-engagement contrastive learning, resulting in improved codebook utilization, reduced collisions, and infused collaborative signal. Second, we use continued pretraining (CPT) to ground the introduced SID tokens before supervised fine-tuning (SFT) on user interaction sequences. Third, we make SnapLGR serving practical through TensorRT-LLM CUDA-backed beam search and a decentralized worker-loop architecture. In a live A/B test, the launched system increased View Time by 0.37%, Time Spent by 0.09%, Deep Sessions by 0.18%, and Deep Sessions Unique User by 0.11% relative to the existing TIGER-style generative retrieval baseline. We then decompose this offline gap under a fixed tokenizer and quantify the gains due to model architecture, scaling, and pretraining. Overall, our deployment shows that successful production SnapLGR requires joint design across representation learning, vocabulary grounding, and efficient training and serving.

cs.IR↗
arXiv2026

Diffusion-aided Task-oriented Semantic Communications with Model Inversion Attack

Semantic communication enhances transmission efficiency by conveying semantic information rather than raw input symbol sequences. Task-oriented semantic communication further aims to retain only task-specific information, thereby achieving greater bandwidth savings. However, these neural-network-based communication systems are vulnerable to model inversion attacks, in which adversaries attempt to recover sensitive input information from intercepted semantic features. The key challenge is therefore to preserve privacy while maintaining task accuracy and robustness. We consider a task-confidential setting in which the adversary attempts to reconstruct the original input from intercepted features without knowing the legitimate receiver's task or model. Although PSNR and SSIM are commonly used to assess reconstruction quality, we find that an external classifier can still perform the legitimate receiver's task with nontrivial accuracy on reconstructions with low PSNR or SSIM, indicating that these reconstructions still contain task-level semantic leakage. We therefore propose DiffSem, which splits the diffusion process between controlled transmitter-side self-noising and matched receiver-side reverse denoising. Experiments on the MNIST, CIFAR-10, and CelebA datasets show that DiffSem improves the legitimate receiver's task accuracy without increasing either the transmitted feature size or information leakage.

cs.CR↗
arXiv2026

ATOD: Annealed Turn-Aware On-Policy Distillation for Multi-Turn Agentic Tasks

Training small language-model agents for long-horizon interactive tasks requires both fast imitation and reward-driven improvement. On-policy distillation (OPD) provides dense teacher guidance and typically improves rapidly in the early stage, but its gains saturate once the student approaches the teacher, limiting the final performance ceiling. Reinforcement learning (RL) directly optimizes environment rewards and encourages exploratory improvement toward a higher reward-defined ceiling, but sparse and delayed feedback makes early-stage learning much less efficient than OPD. In this paper, we propose ATOD (Annealed Turn-aware On-policy Distillation), a hybrid online distillation algorithm that explicitly exploits this complementarity. (1) ATOD uses an annealed OPD-RL schedule: OPD dominates early training to approach teacher-level behavior, while RL is gradually strengthened to drive reward-based exploration. (2) ATOD introduces Turn-level Disagreement-Uncertainty Reweighting (T-DUR), which softly gates the distillation sig- nal to prioritize turns with high disagreement or uncertainty in long trajectories. Experiments on ALFWorld, WebShop, and Search-QA show that ATOD consistently outperforms competing post-training baselines: across the three student sizes, ATOD improves average success rate by 4.16 points over OPD and 23.62 points over GRPO, while surpassing the corresponding teacher models by 2.16 points.

cs.AI↗
arXiv2026

Kimi K2.5: Visual Agentic Intelligence

We introduce Kimi K2.5, an open-source multimodal agentic model designed to advance general agentic intelligence. K2.5 emphasizes the joint optimization of text and vision so that two modalities enhance each other. This includes a series of techniques such as joint text-vision pre-training, zero-vision SFT, and joint text-vision reinforcement learning. Building on this multimodal foundation, K2.5 introduces Agent Swarm, a self-directed parallel agent orchestration framework that dynamically decomposes complex tasks into heterogeneous sub-problems and executes them concurrently. Extensive evaluations show that Kimi K2.5 achieves state-of-the-art results across various domains including coding, vision, reasoning, and agentic tasks. Agent Swarm also reduces latency by up to $4.5\times$ over single-agent baselines. We release the post-trained Kimi K2.5 model checkpoint to facilitate future research and real-world applications of agentic intelligence.

cs.CL↗
arXiv2026

Estimating GHG Emissions from AI Use: Framework for Corporate-Level Measurement

Electricity demand from data centers is expected to grow from roughly 5% of U.S. consumption in 2025 to between 9% and 17% by 2030, and corporate artificial intelligence (AI) use is following a similar trajectory, spanning employee productivity assistants, direct access to large language models (LLMs), and AI features embedded in enterprise software. AI emissions today are a small share of footprints for many enterprises, but that share is unlikely to remain small for long. Without reasonable estimates, companies cannot set reduction targets or identify effective decarbonization levers as emissions grow. Companies, regulators, and auditors are asking for emissions estimates that withstand scrutiny, but no widely accepted methodology exists today. Published per-query estimates can differ by several orders of magnitude depending on what is counted, which provider is measured, and what assumptions are made about electricity use and the grid mix. This white paper proposes a standardized framework for corporate-level AI emissions accounting. The framework is designed to be defensible with current data constraints, tiered to meet companies where their data are, transparent about its assumptions, updatable as provider disclosure matures, and built for action rather than disclosure alone. Since AI emissions accounting is still nascent, it has the opportunity to design for actionability from the outset, so that measurement incentivizes responsible choices during AI's rapid buildout.

physics.soc-ph↗
arXiv2026

MemTX: Transactional Belief Commit for Stateful Agent Memory

LLM agents increasingly coordinate through persistent shared memory: one agent's write becomes another agent's premise, and eventually a tool call with real side effects. Current agent memory systems treat every accepted write as immediately actionable truth, so a polluted tool result, a stale update, or a teammate's half-finished note can silently drive an irreversible action. We argue that a memory write is not a belief commit. We present MemTX, a transactional belief-commit protocol. Each record carries evidence, permissions, provenance, and validity. Writes are staged inside snapshot-isolated transactions and admitted by a validate-and-commit pipeline, irreversible tool calls are gated on in-flight belief state, and retracting a belief triggers typed cascading repair of its derived records and tool side effects. Two invariants, action-safety gating and cascade-repair completeness, are machine-checked by property-based testing and bounded exhaustive enumeration of 5.5 million protocol states, with zero violations. Across five backbones from three model families, MemTX leads all eight baselines with paired-McNemar significance on four backbones and statistically ties the best baseline on the fifth and strongest, while remaining the only method with zero downstream harm on every backbone. Backbone capability does not substitute for commit discipline.

cs.AI↗
arXiv2026

Strategic Plan for Neutral Atom Quantum Computation

We present a strategic plan for neutral atom quantum computation, bringing together hardware development and theory advancements to achieve the goal of practical quantum advantage. The concept of practical quantum advantage is defined, along with how to verify claims of advantage, and approaches to designing quantum algorithms that deliver practical advantage. Future directions for neutral atom quantum processor hardware are described: scaling-up system size, Qubit encodings and atomic platforms, going further below threshold with neutral-atom logical-qubit performance, continuous reloading of qubits, and fast readout. We also explore opportunities for scalable integrated photonic control technologies. Alongside hardware advancements, new developments in quantum error correction and compilation of quantum circuits are proposed. Finally, we examine the opportunity of networking multiple neutral atom quantum processors together to perform distributed quantum computing and overcome possible limitations of a single system.

quant-ph↗
arXiv2026

AgentCompass: A Unified Evaluation Infrastructure for Agent Capabilities

As Large Language Models (LLMs) evolve into autonomous agents, the need for unified evaluation infrastructure becomes critical. However, current evaluation pipelines remain highly fragmented and tightly coupled, hindering reproducibility and causing redundant engineering. To address this, we introduce AgentCompass, an open-source, lightweight, and extensible infrastructure for evaluating LLM-based agents. AgentCompass organizes the evaluation process around three independent components, namely Benchmark, Harness, and Environment, thereby enabling flexible configurations without requiring the reimplementation of complex execution logic. Furthermore, it features a fault-tolerant asynchronous runtime and comprehensive trajectory analysis tools to transparently diagnose nuanced failure modes like reward-hacking. Natively supporting over 20 benchmarks across five capability dimensions, AgentCompass provides the community with a scalable and reproducible infrastructure for advancing agent research.

cs.AI↗
arXiv2026

Resonance-enhanced integrated acousto-optic beam steering

Optical beam steering is a key technology for free-space optical communication, sensing, and imaging. Mechanical beam steering systems suffer from limited scanning speed and bulky form factors, while existing solid-state solutions rely on pixelated synthetic aperture that requires complex fabrication and control architectures. Integrated acousto-optic beam steering (AOBS) is an emerging technology that enables continuous one-dimensional beam steering using integrated acoustic transducers and fixed-wavelength laser sources. Here, we integrate AOBS with an optical ring resonator on the same thin-film lithium niobate (TFLN) platform to significantly enhance beam steering efficiency and system functionality. The resulting device achieves a resonance-enhanced beam steering efficiency of up to $26\%$ and a field of view of $18^\circ$. Moreover, by leveraging integrated electro-optic control, we dynamically lock the ring-resonator's resonance to a chirped laser frequency, enabling frequency-modulated continuous-wave (FMCW) LiDAR operation. By combining lithium niobate's piezoelectric and electro-optic properties, this work establishes a compact, efficient, and scalable beam-steering platform with co-integrated acousto-optic modulation and electro-optic control for multifunctional applications.

physics.app-ph↗
arXiv2026

GatedLinear: Adaptive Routing of Complementary Linear Bases for Time Series Forecasting

Time series forecasting requires models to capture diverse, often mutually exclusive, temporal dynamics, from smooth trend continuation to nonstationary drift and strict phase-aligned recurrence. While recent deep learning models have improved accuracy, they typically force these diverse patterns through a single computational backbone governed by fixed algorithmic inductive biases (e.g., self-attention or spectral filtering). This single-mechanism approach often struggles with the profound heterogeneity of real-world series, where different variables and forecast horizons necessitate fundamentally different predictive treatments. To address this, we propose GatedLinear: a lightweight framework that frames forecasting as the adaptive routing of complementary linear bases. GatedLinear leverages a pool of three specialized mechanisms: a global trend-seasonal basis for smooth projection, a difference-based incremental basis for nonstationary drift, and a phase-aligned recurrence basis for explicit cyclic reuse. To dynamically orchestrate these distinct behaviors, we introduce a Tri-Factorized Fusion Gate that disentangles routing decisions into channel-specific preferences, horizon-aware offsets, and phase-indexed biases derived from known future time marks. This design allows the model to perform highly granular, point-wise soft routing across different predictive regimes without stacking computationally heavy neural modules. Experiments on standard benchmarks show that our method achieves state-of-the-art or highly competitive accuracy against recent complex foundational models, while offering explicitly interpretable routing patterns and operating with a substantially smaller parameter footprint.

cs.LG↗
arXiv2026

Plasma-state metasurfaces for ultra-intensive field manipulation

High-power lasers offer ultrahigh intensities for plasma interactions, but they lack advanced techniques to control the properties of the fields, because no optical elements could withstand their high intensities. The vibrant field of metasurfaces has transformed modern optics by enabling unprecedented control over light at subwavelength through deliberate design. However, metasurfaces have traditionally been limited to solid-state materials and low light intensities. Extending the sophisticated capabilities of metasurfaces from solids into the plasma realm would open new horizons for high-field science. Here, we experimentally demonstrate plasma-state metasurfaces (PSMs) through the photonic spin Hall effect and stable-propagating vortex beam generation irradiated by intense light. Time-resolved pump-probe measurements reveal that the functionality of PSMs can persist for several picoseconds, making them suitable for controlling ultra-intense femtosecond lasers, even in state-of-the-art multi-petawatt systems. Harnessing the powerful toolkit of metasurfaces, this approach holds the promise to revolutionize our ability to manipulate the amplitude, phase, polarization, and wavefront of high-power lasers during their pulse duration. It also opens new possibilities for innovative applications in laser-plasma interactions such as compact particle acceleration and novel radiation sources.

physics.plasm-ph↗
arXiv2026

VLMEvalKit: An Open-Source Toolkit for Evaluating Large Multi-Modality Models

We present VLMEvalKit: an open-source toolkit for evaluating large multi-modality models based on PyTorch. The toolkit aims to provide a user-friendly and comprehensive framework for researchers and developers to evaluate existing multi-modality models and publish \textbf{reproducible} evaluation results. In VLMEvalKit, we implement over 450+ large multi-modality model configurations, including both proprietary APIs and open-source models, and support 330+ benchmarks across diverse multi-modal benchmarks. By implementing a single interface, new models can be easily added to the toolkit, while the toolkit automatically handles the remaining workloads, including data preparation, distributed inference, prediction post-processing, and metric calculation. VLMEvalKit has also evolved to a broader evaluation suite spanning video/audio, document understanding, GUI grounding, spatial reasoning, safety, scientific reasoning, and multi-turn dialogue. Based on the evaluation results obtained with the toolkit, we host the OpenVLM Leaderboard, a comprehensive leaderboard to track the progress of multi-modality learning research. The toolkit is released on https://github.com/open-compass/VLMEvalKit and is actively maintained.

cs.CV↗
arXiv2026

changepointGA: An R package for Fast Changepoint Detection via Genetic Algorithm

Detecting changepoints in a time series of length $N$ entails evaluating up to $2^{N-1}$ possible changepoint models, making exhaustive enumeration computationally infeasible. Genetic algorithms (GAs) provide a stochastic way to identify the structural changes: a population of candidate models evolves via selection, crossover, and mutation operators until it converges on one changepoint model that balances the goodness-of-fit with parsimony. The R package changepointGA encodes each candidate model as an integer chromosome vector and supports both the basic single-population model GA and the island model GA. Parallel computing is implemented on multi-core hardware to further accelerate computation. Users may supply custom fitness functions or genetic operators, while a user-friendly wrapper streamlines routine analyses. Extensive simulations demonstrate that our package runs significantly faster than binary-encoded GA alternatives. Additionally, this package can simultaneously locate changepoints and estimate their effects, as well as other model parameters and any integer-valued hyperparameters. Applications to array-based comparative genomic hybridization data and a century-long temperature series further highlight the package's value in biological and climate research.

stat.CO↗
arXiv2026

Snatcher: Apple Find My Network Exposes Your Lost Devices To Strangers

Apple's Find My network connects nearly one billion devices to locate missing property via Bluetooth Low Energy (BLE). This paper reveals that insecure BLE advertisements and design tradeoffs allow unauthorized discovery and physical theft of lost Apple devices. We develop Snatcher, an attack and analysis framework implemented fully on Android smartphones without specialized hardware. Snatcher identifies vulnerabilities in unencrypted BLE advertisements, unauthenticated acoustic triggers, and slow MAC address randomization. Through three levels - sound-based direction finding, RSSI-IMU sensor-fusion navigation, and spatial-temporal clustering - our Android-based platform physically tracks and locates lost Apple accessories and devices in real-world tests. Our results highlight a crucial conflict between privacy protection, anti-stalking design, and physical security, urging Apple to strengthen Find My defenses.

cs.CR↗
arXiv2026

OpenCompass: A Universal Evaluation Platform for Large Language Models

In recent years, the field of artificial intelligence has undergone a paradigm shift from task-specific small-scale models to general-purpose large language models (LLMs). With the rapid iteration of LLMs, objective, quantitative, and comprehensive evaluation of their capabilities has become a critical link in advancing technological development. Currently, the mainstream static benchmark dataset-based evaluation methods face challenges such as the diversity of task types, inconsistent evaluation criteria, and fragmentation of data and processing workflows, making it difficult to efficiently conduct cross-domain and large-scale model evaluation. To address the aforementioned issues, this paper proposes and open-sources OpenCompass, a one-stop, scalable, and high-concurrency-supported general-purpose LLM evaluation platform. Adhering to the design philosophy of modularization and component decoupling, the platform boasts three core advantages: high compatibility, flexibility, and high concurrency. The core architecture of OpenCompass comprises five key components: the Configuration System, Task Partitioning Module, Execution and Scheduling Module, Task Execution Unit, and Result Visualization Module. Its workflow provides rule-based, LLM-as-a-Judge, and cascaded evaluators to adapt to the requirements of different task scenarios. Supporting mainstream benchmark datasets across multiple domains, including knowledge, reasoning, computation, science, language, code, etc., the platform offers a unified and efficient LLM evaluation tool for both academia and industry, facilitating the accurate identification of strengths and weaknesses of LLMs as well as their subsequent optimization.

cs.CL↗