Jyun-Ze Tang
Scholar

Jyun-Ze Tang

Google Scholar ID: SXNQEP4AAAAJ
Inventec Corporation
DiffusionAnomaly DetectionImage SynthesisImage Inpainting
Homepage↗Google Scholar↗
Citations & Impact
All-time
Citations
0
 
H-index
0
 
i10-index
0
 
Publications
1
 
Co-authors
0
 
Contact
Emailcol360785@gmail.comGitHubOpen ↗
Publications
1 items
LSSGen: Leveraging Latent Space Scaling in Flow and Diffusion for Efficient Text to Image Generation
2025
Cited
0
Resume (English only)
Academic Achievements
  • Published several technical notes or paper summaries on applications of computer graphics and deep learning, specifically including:
  • - Discussion on common issues with diffusion noise schedules and sample steps
  • - Methods for parameter sharing between different modules in PyTorch Lightning
  • - Manga inpainting and restoration based on latent diffusion models
  • - Practical use of OneCycleLR scheduler in PyTorch Lightning
  • - Implementation of image similarity and semantic search using the CLIP model and Milvus DB
Research Experience
  • Research areas include analysis of diffusion model noise schedules and sample steps, parameter sharing techniques within the PyTorch Lightning framework, a project on manga inpainting and restoration based on latent diffusion models, and implementing image similarity and semantic search using the CLIP model with Milvus vector database.
Background
  • Member of NTUST CSIE Computer Graphics Lab, focusing on research in computer graphics and related fields.
Miscellany
  • Location: Taiwan
  • Social Media/Contact: GitHub, Email
Co-authors
0 total
Co-authors: 0 (list not available)