arXiv subjects

Zhijing Li

Publications and source records attributed to Zhijing Li.

12 recordsLinked to original sources

arXiv2026

GR2 Technical Report

Industrial recommendation systems serve billions of users through a multi-stage funnel -- retrieval, early-stage ranking, and re-ranking -- where the final re-ranking step disproportionately shapes user engagement and downstream performance, particularly for carousel and grid display formats. Despite growing enthusiasm for Large Language Models (LLMs) in recommendation, three gaps hinder industrial adoption: (1) most efforts target retrieval and ranking, leaving re-ranking -- the stage closest to the final user experience -- largely underexplored; (2) LLMs are typically deployed zero-shot or via supervised fine-tuning, underutilizing the reasoning capabilities unlocked by reinforcement learning (RL) on verifiable rewards; (3) deployed catalogs index billions of items with non-semantic identifiers that lie outside any base-LLM vocabulary. We present GR2 (Generative Reasoning Re-Ranker), an end-to-end framework that combines (i) mid-training on semantic IDs produced by a tokenizer with >=99% uniqueness, (ii) reasoning-trace distilled from a stronger teacher via targeted prompting and rejection sampling, and (iii) RL with verifiable rewards purpose-built for re-ranking. To make GR2 resource-viable, we further (iv) introduce a context compressor that amortizes training cost, On-Policy Distillation (OPD) as a scalable alternative to SFT -- which we find collapses at industrial scale -- and reasoning distillation for low-latency serving. GR2 delivers +18.7% R@1, +7.1% R@3, and +9.6% N@3 over legacy baselines on industrial-scale traffic. We further find that reward design is critical in re-ranking: LLMs often hack rewards by preserving the incoming order or exploiting position bias, motivating conditional verifiable rewards as essential industrial components.

cs.IR↗

arXiv2026

GR2: Generative Reasoning Re-ranker

Recent studies increasingly explore Large Language Models (LLMs) as a new paradigm for recommendation systems due to their scalability and world knowledge. However, existing work has three key limitations: (1) most efforts focus on retrieval and ranking, while the reranking phase, critical for refining final recommendations, is largely overlooked; (2) LLMs are typically used in zero-shot or supervised fine-tuning settings, leaving their reasoning abilities, especially those enhanced through reinforcement learning (RL) and high-quality reasoning data, underexploited; (3) items are commonly represented by non-semantic IDs, creating major scalability challenges in industrial systems with billions of identifiers. To address these gaps, we propose the Generative Reasoning Reranker (GR2), an end-to-end framework with a three-stage training pipeline tailored for reranking. First, a pretrained LLM is mid-trained on semantic IDs encoded from non-semantic IDs via a tokenizer achieving $\ge$99% uniqueness. Next, a stronger larger-scale LLM generates high-quality reasoning traces through carefully designed prompting and rejection sampling, which are used for supervised fine-tuning to impart foundational reasoning skills. Finally, we apply Decoupled Clip and Dynamic sAmpling Policy Optimization (DAPO), enabling scalable RL supervision with verifiable rewards designed specifically for reranking. Experiments on two real-world datasets demonstrate GR2's effectiveness: it surpasses the state-of-the-art OneRec-Think by 2.4% in Recall@5 and 1.3% in NDCG@5. Ablations confirm that advanced reasoning traces yield substantial gains across metrics. We further find that RL reward design is crucial in reranking: LLMs tend to exploit reward hacking by preserving item order, motivating conditional verifiable rewards to mitigate this behavior and optimize reranking performance.

cs.IR↗

arXiv2026

MicLog: Towards Accurate and Efficient LLM-based Log Parsing via Progressive Meta In-Context Learning

Log parsing converts semi-structured logs into structured templates, forming a critical foundation for downstream analysis. Traditional syntax and semantic-based parsers often struggle with semantic variations in evolving logs and data scarcity stemming from their limited domain coverage. Recent large language model (LLM)-based parsers leverage in-context learning (ICL) to extract semantics from examples, demonstrating superior accuracy. However, LLM-based parsers face two main challenges: 1) underutilization of ICL capabilities, particularly in dynamic example selection and cross-domain generalization, leading to inconsistent performance; 2) time-consuming and costly LLM querying. To address these challenges, we present MicLog, the first progressive meta in-context learning (ProgMeta-ICL) log parsing framework that combines meta-learning with ICL on small open-source LLMs (i.e., Qwen-2.5-3B). Specifically, MicLog: i) enhances LLMs' ICL capability through a zero-shot to k-shot ProgMeta-ICL paradigm, employing weighted DBSCAN candidate sampling and enhanced BM25 demonstration selection; ii) accelerates parsing via a multi-level pre-query cache that dynamically matches and refines recently parsed templates. Evaluated on Loghub-2.0, MicLog achieves 10.3% higher parsing accuracy than the state-of-the-art parser while reducing parsing time by 42.4%.

cs.SE↗

arXiv2025

External Large Foundation Model: How to Efficiently Serve Trillions of Parameters for Online Ads Recommendation

Ads recommendation is a prominent service of online advertising systems and has been actively studied. Recent studies indicate that scaling-up and advanced design of the recommendation model can bring significant performance improvement. However, with a larger model scale, such prior studies have a significantly increasing gap from industry as they often neglect two fundamental challenges in industrial-scale applications. First, training and inference budgets are restricted for the model to be served, exceeding which may incur latency and impair user experience. Second, large-volume data arrive in a streaming mode with data distributions dynamically shifting, as new users/ads join and existing users/ads leave the system. We propose the External Large Foundation Model (ExFM) framework to address the overlooked challenges. Specifically, we develop external distillation and a data augmentation system (DAS) to control the computational cost of training/inference while maintaining high performance. We design the teacher in a way like a foundation model (FM) that can serve multiple students as vertical models (VMs) to amortize its building cost. We propose Auxiliary Head and Student Adapter to mitigate the data distribution gap between FM and VMs caused by the streaming data issue. Comprehensive experiments on internal industrial-scale applications and public datasets demonstrate significant performance gain by ExFM.

cs.IR↗

arXiv2023

A Distributed Data-Parallel PyTorch Implementation of the Distributed Shampoo Optimizer for Training Neural Networks At-Scale

Shampoo is an online and stochastic optimization algorithm belonging to the AdaGrad family of methods for training neural networks. It constructs a block-diagonal preconditioner where each block consists of a coarse Kronecker product approximation to full-matrix AdaGrad for each parameter of the neural network. In this work, we provide a complete description of the algorithm as well as the performance optimizations that our implementation leverages to train deep networks at-scale in PyTorch. Our implementation enables fast multi-GPU distributed data-parallel training by distributing the memory and computation associated with blocks of each parameter via PyTorch's DTensor data structure and performing an AllGather primitive on the computed search directions at each iteration. This major performance enhancement enables us to achieve at most a 10% performance reduction in per-step wall-clock time compared against standard diagonal-scaling-based adaptive gradient methods. We validate our implementation by performing an ablation study on training ImageNet ResNet50, demonstrating Shampoo's superiority over standard training recipes with minimal hyperparameter tuning.

cs.LG↗

arXiv2023

Hue: A User-Adaptive Parser for Hybrid Logs

Log parsing, which extracts log templates from semi-structured logs and produces structured logs, is the first and the most critical step in automated log analysis. While existing log parsers have achieved decent results, they suffer from two major limitations by design. First, they do not natively support hybrid logs that consist of both single-line logs and multi-line logs (\eg Java Exception and Hadoop Counters). Second, they fall short in integrating domain knowledge in parsing, making it hard to identify ambiguous tokens in logs. This paper defines a new research problem, \textit{hybrid log parsing}, as a superset of traditional log parsing tasks, and proposes \textit{Hue}, the first attempt for hybrid log parsing via a user-adaptive manner. Specifically, Hue converts each log message to a sequence of special wildcards using a key casting table and determines the log types via line aggregating and pattern extracting. In addition, Hue can effectively utilize user feedback via a novel merge-reject strategy, making it possible to quickly adapt to complex and changing log templates. We evaluated Hue on three hybrid log datasets and sixteen widely-used single-line log datasets (\ie Loghub). The results show that Hue achieves an average grouping accuracy of 0.845 on hybrid logs, which largely outperforms the best results (0.563 on average) obtained by existing parsers. Hue also exhibits SOTA performance on single-line log datasets. Furthermore, Hue has been successfully deployed in a real production environment for daily hybrid log parsing.

cs.SE↗

arXiv2022

Compiler-Driven Simulation of Reconfigurable Hardware Accelerators

As customized accelerator design has become increasingly popular to keep up with the demand for high performance computing, it poses challenges for modern simulator design to adapt to such a large variety of accelerators. Existing simulators tend to two extremes: low-level and general approaches, such as RTL simulation, that can model any hardware but require substantial effort and long execution times; and higher-level application-specific models that can be much faster and easier to use but require one-off engineering effort. This work proposes a compiler-driven simulation workflow that can model configurable hardware accelerator. The key idea is to separate structure representation from simulation by developing an intermediate language that can flexibly represent a wide variety of hardware constructs. We design the Event Queue (EQueue) dialect of MLIR, a dialect that can model arbitrary hardware accelerators with explicit data movement and distributed event-based control; we also implement a generic simulation engine to model EQueue programs with hybrid MLIR dialects representing different abstraction levels. We demonstrate two case studies of EQueue-implemented accelerators: the systolic array of convolution and SIMD processors in a modern FPGA. In the former we show EQueue simulation is as accurate as a state-of-the-art simulator, while offering higher extensibility and lower iteration cost via compiler passes. In the latter we demonstrate our simulation flow can guide designer efficiently improve their design using visualizable simulation outputs.

cs.PL↗

arXiv2021

A Compiler Infrastructure for Accelerator Generators

We present Calyx, a new intermediate language (IL) for compiling high-level programs into hardware designs. Calyx combines a hardware-like structural language with a software-like control flow representation with loops and conditionals. This split representation enables a new class of hardware-focused optimizations that require both structural and control flow information which are crucial for high-level programming models for hardware design. The Calyx compiler lowers control flow constructs using finite-state machines and generates synthesizable hardware descriptions. We have implemented Calyx in an optimizing compiler that translates high-level programs to hardware. We demonstrate Calyx using two DSL-to-RTL compilers, a systolic array generator and one for a recent imperative accelerator language, and compare them to equivalent designs generated using high-level synthesis (HLS). The systolic arrays are $4.6\times$ faster and $1.1\times$ larger on average than HLS implementations, and the HLS-like imperative language compiler is within a few factors of a highly optimized commercial HLS toolchain. We also describe three optimizations implemented in the Calyx compiler.

cs.PL↗

arXiv2020

Predictable Accelerator Design with Time-Sensitive Affine Types

Field-programmable gate arrays (FPGAs) provide an opportunity to co-design applications with hardware accelerators, yet they remain difficult to program. High-level synthesis (HLS) tools promise to raise the level of abstraction by compiling C or C++ to accelerator designs. Repurposing legacy software languages, however, requires complex heuristics to map imperative code onto hardware structures. We find that the black-box heuristics in HLS can be unpredictable: changing parameters in the program that should improve performance can counterintuitively yield slower and larger designs. This paper proposes a type system that restricts HLS to programs that can predictably compile to hardware accelerators. The key idea is to model consumable hardware resources with a time-sensitive affine type system that prevents simultaneous uses of the same hardware structure. We implement the type system in Dahlia, a language that compiles to HLS C++, and show that it can reduce the size of HLS parameter spaces while accepting Pareto-optimal designs.

cs.PL↗

arXiv2020

Optimizing JPEG Quantization for Classification Networks

Deep learning for computer vision depends on lossy image compression: it reduces the storage required for training and test data and lowers transfer costs in deployment. Mainstream datasets and imaging pipelines all rely on standard JPEG compression. In JPEG, the degree of quantization of frequency coefficients controls the lossiness: an 8 by 8 quantization table (Q-table) decides both the quality of the encoded image and the compression ratio. While a long history of work has sought better Q-tables, existing work either seeks to minimize image distortion or to optimize for models of the human visual system. This work asks whether JPEG Q-tables exist that are "better" for specific vision networks and can offer better quality--size trade-offs than ones designed for human perception or minimal distortion. We reconstruct an ImageNet test set with higher resolution to explore the effect of JPEG compression under novel Q-tables. We attempt several approaches to tune a Q-table for a vision task. We find that a simple sorted random sampling method can exceed the performance of the standard JPEG Q-table. We also use hyper-parameter tuning techniques including bounded random search, Bayesian optimization, and composite heuristic optimization methods. The new Q-tables we obtained can improve the compression rate by 10% to 200% when the accuracy is fixed, or improve accuracy up to $2\%$ at the same compression rate.

cs.CV↗

arXiv2020

Et Tu Alexa? When Commodity WiFi Devices Turn into Adversarial Motion Sensors

Our work demonstrates a new set of silent reconnaissance attacks, which leverages the presence of commodity WiFi devices to track users inside private homes and offices, without compromising any WiFi network, data packets, or devices. We show that just by sniffing existing WiFi signals, an adversary can accurately detect and track movements of users inside a building. This is made possible by our new signal model that links together human motion near WiFi transmitters and variance of multipath signal propagation seen by the attacker sniffer outside of the property. The resulting attacks are cheap, highly effective, and yet difficult to detect. We implement the attack using a single commodity smartphone, deploy it in 11 real-world offices and residential apartments, and show it is highly effective. Finally, we evaluate potential defenses, and propose a practical and effective defense based on AP signal obfuscation.

cs.CR↗

arXiv2019

Understanding the Effectiveness of Ultrasonic Microphone Jammer

Recent works have explained the principle of using ultrasonic transmissions to jam nearby microphones. These signals are inaudible to nearby users, but leverage "hardware nonlinearity" to induce a jamming signal inside microphones that disrupts voice recordings. This has great implications on audio privacy protection. In this work, we gain a deeper understanding on the effectiveness of ultrasonic jammer under practical scenarios, with the goal of disabling both visible and hidden microphones in the surrounding area. We first experiment with existing jammer designs (both commercial products and that proposed by recent papers), and find that they all offer limited angular coverage, and can only target microphones in a particular direction. We overcome this limitation by building a circular transducer array as a wearable bracelet. It emits ultrasonic signals simultaneously from many directions, targeting surrounding microphones without needing to point at any. More importantly, as the bracelet moves with the wearer, its motion increases jamming coverage and diminishes blind spots (the fundamental problem facing any transducer array). We evaluate the jammer bracelet under practical scenarios, confirming that it can effectively disrupt visible and hidden microphones in the surrounding areas, preventing recognition of recorded speech. We also identify limitations and areas for improvement.

cs.CR↗