arXiv subjects

Zelong Zhao

Publications and source records attributed to Zelong Zhao.

6 recordsLinked to original sources

arXiv2026

Argus: A General-Purpose Agentic Reasoning Runtime for Long-Horizon Tasks

Long-horizon reasoning requires an agentic runtime that can persist when evidence supports its current approach and pivot when measurements reveal failure, hidden constraints, or a misspecified objective. We present Argus, a persistent, self-evolving runtime in which Manager, Planner, Engineer, and Reviewer execute bounded missions over durable project state. Argus separates stable user intent from operational objectives, constraints, and verification criteria, and admits memories, skills, procedures, verifiers, routing decisions, and rejected routes only after role-owned review and, when available, task-native verification. Model weights remain fixed; self-evolution occurs through persistent runtime state and control policy, with autonomous execution between operator-owned escalation points. Across seven GPT-5.5 benchmark arenas, Argus achieves about 78% on SWE-Bench Pro versus 59% for Direct Copilot while using 1.41 times the aggregate tokens. After verification-gated self-evolution, mature SWE-Bench waves use 21% fewer solve-input tokens and 15% less active workflow time per task than startup waves, while recording 34 verifier recoveries and 22 strict review-loop rescues. Argus also reaches 76.8% on AARRI-Bench and a 28.0-point gap on mathematical data synthesis, with competitive GPU-kernel and language-model-training results. Beyond benchmarks, an optimized RWKV6 kernel was merged upstream; a multi-day mathematics campaign retained falsified routes and proof-backed frontier updates; and six paper pipelines completed 254 missions with 16 stage rollbacks. These results show that a fixed-weight, self-evolving harness can revise, recover, and accumulate verified approaches while producing structured trajectories for future supervised and reinforcement learning.

cs.AI↗

arXiv2026

LLM-Based User Personas for Recommendations at Scale

Large Language Models (LLMs) offer unprecedented potential for enhancing recommendation systems through their world knowledge and reasoning capabilities. However, existing approaches often rely on structured IDs or offline processing, limiting semantic richness, real-time adaptability, and user-facing interpretability. In this paper, we introduce a novel framework that enables real-time generation of LLM-based user interest personas for a large-scale commercial video recommendation platform. Our method generates natural-language user interest personas that address the exploitation-exploration trade-off by combining the summarization of existing interests with novel topics, directly during serving. To overcome the computational challenges of online LLM inference at a billion-user scale, we design a cost-efficient architecture leveraging knowledge distillation, asynchronous inference, and input optimization via semantically clustered video representations. Extensive offline evaluations, user studies, and live A/B tests demonstrate significant improvements in viewer value. This work bridges the gap between high-level semantic understanding and industrial-scale recommendation, paving the way for more dynamic, explainable, and satisfying personalized experiences.

cs.IR↗

arXiv2023

A language-inspired machine learning approach for solving strongly correlated problems with dynamical mean-field theory

We present SCALINN -- Strongly Correlated Approach with Language Inspired Neural Network -- as a method for solving the Anderson impurity model and reducing the computational cost of dynamical mean-field theory calculations. Inspired by the success of generative Transformer networks in natural language processing, SCALINN utilizes an in-house modified Transformer network in order to learn correlated Matsubara Green's functions, which act as solutions to the impurity model. This is achieved by providing the network with low-cost Matsubara Green's functions, thereby overcoming the computational cost of high accuracy solutions. Across different temperatures and interaction strengths, the performance of SCALINN is demonstrated in both physical observables (spectral function, Matsubara Green's functions, quasi-particle weight), and the mean squared error cost values of the neural network, showcasing the network's ability to accelerate Green's function based calculations of correlated materials.

cond-mat.str-el↗

arXiv2022

Computational prediction of samarium hydride at megabar pressure

Samarium hydrides, belonging to the broad class of lanthanide polyhydrides, have yet to be experimentally tested at high pressure. In this study, we use random structure searches to explore multiple possible stoichiometries and propose SmH$_2$ with a layered hexagonal structure in the $P6/mmm$ space group and SmH$_6$ with hydrogen clathrate structures in the Im-3m space group as theoretically stable phases of samarium hydrides at a wide range of pressures centered around $200$ GPa. We further combine the first-principles methods of density functional theory and dynamical mean-field theory to explore many-body correlations in samarium hydrides, reporting electron and phonon dispersions and densities of states, and also evaluate the electron-phonon driven superconductivity to investigate low critical temperatures at $200$ GPa.

cond-mat.str-el↗

arXiv2021

High temperature superconductivity in the lanthanide hydrides at extreme pressures

Hydrogen-rich superhydrides are promising high-Tc superconductors, with superconductivity experimentally observed near room temperature, as shown in recently discovered lanthanide superhydrides at very high pressures, e.g. LaH10 at 170 GPa and CeH9 at 150 GPa. Superconductivity is believed to be closely related with the high vibrational modes of the bound hydrogen ions. Here we study the limit of extreme pressures (above 200 GPa) where lanthanide hydrides with large hydrogen content have been reported. We focus on LaH16 and CeH16, two prototype candidates for achieving a large electronic contribution from hydrogen in the electron-phonon coupling. In this work, we propose a first principles calculation platform with the inclusion of many-body corrections to evaluate the detailed physical properties of the Ce-H and La-H systems and to understand the structure, stability and superconductivity of these systems at ultra-high pressure. We provide a practical approach to further investigate conventional superconductivity in hydrogen rich superhydrides. We report that density functional theory provides accurate structure and phonon frequencies, but many-body corrections lead to an increase of the critical temperature, that is associated with spectral weight transfer of the f-states.

cond-mat.supr-con↗

arXiv2021

Computational Materials Discovery for Lanthanide Hydrides at high pressure: predicting High Temperature superconductivity

Hydrogen-rich superhydrides are believed to be very promising high-T$_c$ superconductors, with experimentally observed critical temperatures near room temperature, as shown in recently discovered lanthanide superhydrides at very high pressures, e.g. LaH$_{10}$ at 170 GPa and CeH$_9$ at 150 GPa. With the motivation of discovering new hydrogen-rich high-T$_c$ superconductors at lowest possible pressure, quantitative theoretical predictions are needed. In these promising compounds, superconductivity is mediated by the highly energetic lattice vibrations associated with hydrogen and their interplay with the electronic structure, requiring fine descriptions of the electronic properties, notoriously challenging for correlated $f$ systems. In this work, we propose a first-principles calculation platform with the inclusion of many-body corrections to evaluate the detailed physical properties of the Ce-H system and to understand the structure, stability and superconductivity of CeH$_9$ at high pressure. We report how the prediction of T$_c$ is affected by the hierarchy of many-body corrections, and obtain a compelling increase of T$_c$ at the highest level of theory, which goes in the direction of experimental observations. Our findings shed a significant light on the search for superhydrides in close similarity with atomic hydrogen within a feasible pressure range. We provide a practical platform to further investigate and understand conventional superconductivity in hydrogen rich superhydrides.

cond-mat.supr-con↗