A Stitch in Time Saves Nine: Small VLM is a Precise Guidance for Accelerating Large VLMs

Documenti analoghi

• Dynamic Diffusion Transformer
  di: Zhao, Wangbo, et al.
  Pubblicazione: (2024)
• Dynamic Tuning Towards Parameter and Inference Efficiency for ViT Adaptation
  di: Zhao, Wangbo, et al.
  Pubblicazione: (2024)
• RAPID^3: Tri-Level Reinforced Acceleration Policies for Diffusion Transformer
  di: Zhao, Wangbo, et al.
  Pubblicazione: (2025)
• DyDiT++: Diffusion Transformers with Timestep and Spatial Dynamics for Efficient Visual Generation
  di: Zhao, Wangbo, et al.
  Pubblicazione: (2025)
• A Stitch in Time Saves Nine: Proactive Self-Refinement for Language Models
  di: Han, Jinyi, et al.
  Pubblicazione: (2025)