arXiv subjects

Zuyu Zhang

Publications and source records attributed to Zuyu Zhang.

4 recordsLinked to original sources

arXiv2026

Git4Data: Database-Native Version Control for AI Agents

Large Language Model (LLM) agents increasingly explore many candidate states of relational data in parallel, each of which should remain isolated, reproducible, and auditable, preferably through the same SQL interface used for ordinary data work. Existing tools support this requirement only partially: source-code version control does not scale to large datasets, whereas relational databases manage large data efficiently but rarely expose native branching, comparison, and merging. We present Git4Data, a database-native version-control layer for agentic workflows. Git4Data treats a database as a repository and a table as a versioned object, exposing Git-style operations (snapshot/tag, branch, diff, and merge with explicit conflict-resolution policies) through SQL extensions. Implemented in MatrixOne, a cloud-native relational database, Git4Data leverages immutable object storage and MVCC to make the cost of these operations proportional to the size of the change rather than the size of the data. On the BranchBench agentic branching workloads, Git4Data outperforms DoltDB by up to an order of magnitude. Overall, we believe this work sheds light on how relational databases can better support AI agents through efficient versioning.

cs.DB↗

arXiv2026

ContextPipe: Database-Inspired Context Assembly for Long-Horizon Agents

Long-horizon large language model (LLM) agents require context assembly: the runtime must decide what to include in each prompt, in what order, and when to compact history under a hard context-window budget and a byte-sensitive prompt cache. In production agentic systems, this logic is scattered across prompt builders, ad hoc compaction routines, cache-break workarounds, and per-provider shims. We argue that context assembly is structurally isomorphic to query execution in a relational database: both execute under a hard budget, exploit a tiered cache, and leverage statistics. We adopt this discipline in ContextPipe: a five-phase pipeline (Plan Bind Optimize Execute Feedback) backed by a structured data-source catalog, a deterministic cache-aware optimizer, and an EXPLAIN ANALYZE trace. We show that context in ContextPipe is auditable, replayable, and failure-isolated. A preliminary evaluation using the SWE-bench Pro Qutebrowser subset shows that, compared with the append-only context construction policy, ContextPipe reduces total token volume by 31%, LLM calls by 23%, and response time by 9%, at the cost of a lower KV cache-hit ratio.

cs.AI↗

arXiv2025

Adversarial Data Augmentation for Single Domain Generalization via Lyapunov Exponent-Guided Optimization

Single Domain Generalization (SDG) aims to develop models capable of generalizing to unseen target domains using only one source domain, a task complicated by substantial domain shifts and limited data diversity. Existing SDG approaches primarily rely on data augmentation techniques, which struggle to effectively adapt training dynamics to accommodate large domain shifts. To address this, we propose LEAwareSGD, a novel Lyapunov Exponent (LE)-guided optimization approach inspired by dynamical systems theory. By leveraging LE measurements to modulate the learning rate, LEAwareSGD encourages model training near the edge of chaos, a critical state that optimally balances stability and adaptability. This dynamic adjustment allows the model to explore a wider parameter space and capture more generalizable features, ultimately enhancing the model's generalization capability. Extensive experiments on PACS, OfficeHome, and DomainNet demonstrate that LEAwareSGD yields substantial generalization gains, achieving up to 9.47\% improvement on PACS in low-data regimes. These results underscore the effectiveness of training near the edge of chaos for enhancing model generalization capability in SDG tasks.

cs.CV↗

arXiv2018

Scaling-Up In-Memory Datalog Processing: Observations and Techniques

Recursive query processing has experienced a recent resurgence, as a result of its use in many modern application domains, including data integration, graph analytics, security, program analysis, networking and decision making. Due to the large volumes of data being processed, several research efforts, across multiple communities, have explored how to scale up recursive queries, typically expressed in Datalog. Our experience with these tools indicated that their performance does not translate across domains (e.g., a tool design for large-scale graph analytics does not exhibit the same performance on program-analysis tasks, and vice versa). As a result, we designed and implemented a general-purpose Datalog engine, called RecStep, on top of a parallel single-node relational system. In this paper, we outline the different techniques we use in RecStep, and the contribution of each technique to overall performance. We also present results from a detailed set of experiments comparing RecStep with a number of other Datalog systems using both graph analytics and program-analysis tasks, summarizing pros and cons of existing techniques based on the analysis of our observations. We show that RecStep generally outperforms the state-of-the-art parallel Datalog engines on complex and large-scale Datalog program evaluation, by a 4-6X margin. An additional insight from our work is that we show that it is possible to build a high-performance Datalog system on top of a relational engine, an idea that has been dismissed in past work in this area.

cs.DB↗