Summary

Undergraduate student at Xi'an Jiaotong University, majoring in Computer Science and Technology. Research interests in Multimodal Large Language Models, Embodied Intelligence, and Computer Vision.

Education

• Computer Science and Technology

  Present
  Xi'an Jiaotong University

  Courses: Linear Algebra (99), Advanced Mathematics (95), Data Structures (90), OOP (98), Digital Logic Circuits (90), Probability Theory (89), ICS (96), Computer Graphics (95), Computer Networks (89), Operating Systems (85)

Work Experience

• Researcher

  -
  Built a comprehensive audio-video dataset covering 33 fine-grained categories and 190 instructions. Implemented a decoupled evaluation pipeline combining objective physical metrics with modality-separated MLLM checklist. Deployed evaluation on SOTA models including Wan2.7 and LTX2.

  • NeurIPS 2026 under review
  • Cross-modal physical alignment evaluation
  • Human-machine consistency analysis
• Developer

  -
  Developed electronic control system for multi-DOF robotic arm. Applied PID control for gravity compensation, eliminating steady-state errors and motion jitter.

  • Multi-degree-of-freedom robotic arm control
  • PID gravity compensation algorithm
  • Sensor closed-loop control
• Developer

  -
  Completed lightweight deployment of YOLO models on Ascend NPU. Improved FPS through model conversion and edge-side acceleration optimization.

  • Edge-side inference optimization
  • Model conversion and acceleration
  • Reduced transmission latency

Skills

Publications

• AVE-Compass: Towards Holistic Evaluation for Audio-Video Editing Abilities

  2026
  NeurIPS 2026 (under review)

  A comprehensive benchmark for evaluating audio-video editing capabilities, covering 33 fine-grained categories and 190 instructions with decoupled evaluation pipeline.

  View Publication

Languages

• Chinese

  Native
• English

  CET-4: 619, CET-6: 590

Interests

• Multimodal Large Language Models
• Embodied Intelligence
• Computer Vision