arXiv subjects

Shen Yang

Publications and source records attributed to Shen Yang.

6 recordsLinked to original sources

arXiv2026

CogEvol: Towards Efficient and Reliable Learning Environment Generation

We present CogEvol, a family of models trained specifically for Learning Environment Generation: turning a course brief into a finished learning artifact (structured-JSON slides or self-contained interactive HTML pages) in a single pass. Across 220k production requests, CogEvol completes a slide in a median of 17 seconds and an interactive page in 59, replacing minutes-long multi-turn agent scaffolding. Reliability is enforced rather than hoped for: a production-grounded data pipeline turns real failures into 53,687 verified SFT samples, and a hybrid rule-plus-VLM reward drives GRPO-based RL, hardened after we caught and fixed a reward-hacking episode that produced visually convincing but unplayable games. CogEvol-27B scores 83.7 on slide quality and 63.7 on a 500-case interactive-HTML benchmark with 26.9x fewer parameters than flagship coding models, and, in collaboration with the OpenMAIC team, serves their live production traffic. CogEvol-4B is released openly under the Apache 2.0 license at https://github.com/CogEvol/CogEvol-4B; external flagships are measured on the same suites under the identical harness. Scaffold editing cuts interactive-page generation cost by a further ~76%, and the full stack runs on domestic Ascend accelerators at application-level parity with A800 GPUs, lowering the unit cost of AI-native education at scale.

cs.CL↗

arXiv2025

Addressing Situated Teaching Needs: A Multi-Agent Framework for Automated Slide Adaptation

The adaptation of teaching slides to instructors' situated teaching needs, including pedagogical styles and their students' context, is a critical yet time-consuming task for educators. Through a series of educator interviews, we first identify and systematically categorize the key friction points that impede this adaptation process. Grounded in these findings, we introduce a novel multi-agent framework designed to automate slide adaptation based on high-level instructor specifications. An evaluation involving 16 modification requests across 8 real-world courses validates our approach. The framework's output consistently achieved high scores in intent alignment, content coherence and factual accuracy, and performed on par with baseline methods regarding visual clarity, while also demonstrating appropriate timeliness and a high operational agreement with human experts, achieving an F1 score of 0.89. This work heralds a new paradigm where AI agents handle the logistical burdens of instructional design, liberating educators to focus on the creative and strategic aspects of teaching.

cs.MA↗

arXiv2024

VisionReward: Fine-Grained Multi-Dimensional Human Preference Learning for Image and Video Generation

Visual generative models have achieved remarkable progress in synthesizing photorealistic images and videos, yet aligning their outputs with human preferences across critical dimensions remains a persistent challenge. Though reinforcement learning from human feedback offers promise for preference alignment, existing reward models for visual generation face limitations, including black-box scoring without interpretability and potentially resultant unexpected biases. We present VisionReward, a general framework for learning human visual preferences in both image and video generation. Specifically, we employ a hierarchical visual assessment framework to capture fine-grained human preferences, and leverages linear weighting to enable interpretable preference learning. Furthermore, we propose a multi-dimensional consistent strategy when using VisionReward as a reward model during preference optimization for visual generation. Experiments show that VisionReward can significantly outperform existing image and video reward models on both machine metrics and human evaluation. Notably, VisionReward surpasses VideoScore by 17.2% in preference prediction accuracy, and text-to-video models with VisionReward achieve a 31.6% higher pairwise win rate compared to the same models using VideoScore. All code and datasets are provided at https://github.com/THUDM/VisionReward.

cs.CV↗

arXiv2019

Critical parameters of liquid-gas phase transition in thermal symmetric and asymmetric nuclear matter

The properties of critical parameters and phase diagram structure of liquid-gas phase transition are investigated in thermal symmetric and asymmetric nuclear matter with the covariant density functional (CDF) theory. Although uncertainty remains in predicting the critical parameters such as the critical temperature and pressure from various CDF functionals, several correlations are explored numerically and verified to be approximately linear between them. These correlations become worse when nuclear matter is more isospin asymmetric, resulting mainly from the effects induced by symmetry energy. By looking over the isospin dependence of the critical temperature, the role of the symmetry energy in LG transition properties of asymmetric matter is realized. The change of critical temperature with isospin asymmetry is found to be correlated well with and as a consequence could be constrained by the density slope of symmetry energy at saturation density. Then, the structure of phase diagram of thermal nuclear matter is analyzed carefully. It is revealed that the contribution from symmetry energy dominates the size of liquid-gas phase coexistence area. Moreover, the specific pattern of the phase diagram could be determined by the critical temperature at non-zero isospin asymmetry, illustrated from the correlations of the temperature with pressures at several characteristic points, paving the possible way to further explore the structure of liquid-gas phase diagram of thermal nuclear matter.

nucl-th↗

arXiv2015

Self-consistent tensor effects on nuclear matter system under relativistic Hartree-Fock approach

With the relativistic representation of the nuclear tensor force that is included automatically by the Fock diagrams, we explored the self-consistent tensor effects on the properties of nuclear matter system. The analysis were performed within the density-dependent relativistic Hartree-Fock (DDRHF) theory. The tensor force is found to notably influence the saturation mechanism, the equation of state and the symmetry energy of nuclear matter, as well as the neutron star properties. Without introducing any additional free parameters, the DDRHF approach paves a natural way to reveal the tensor effects on the nuclear matter system.

nucl-th↗

arXiv2015

The Nuclear Tensor Interaction in Covariant Energy Density Functional

The origin of the nuclear tensor interaction in the covariant energy density functional (EDF) is presented in this work, associated with the Fock diagrams of Lorentz scalar and vector couplings. With this newly obtained relativistic formalism of the nuclear tensor interaction, more distinct tensor effects are found in the Fock diagrams of the Lorentz scalar and vector couplings, as compared to the Lorentz pseudo-vector and tensor channels. A unified and self-consistent treatment on both the nuclear tensor and spin-orbit interactions, which dominate the spin-dependent features of the nuclear force, is then achieved by the relativistic models. Moreover, careful analysis on the tensor strengths indicates the reliability of the nuclear tensor interaction in the covariant EDF for exploring the nuclear structure, excitation and decay modes.

nucl-th↗