Skip to content

Latest commit

 

History

History
36 lines (23 loc) · 1.04 KB

README.md

File metadata and controls

36 lines (23 loc) · 1.04 KB

Accelerating T2I-AR with SJD

Implementation of Accelerating Auto-regressive Text-to-Image Generation with Training-free Speculative Jacobi Decoding

Performance

Run

demos

Lumina-mGPT

CUDA_VISIBLE_DEVICES=0 python test_lumina_mgpt.py

Emu3

CUDA_VISIBLE_DEVICES=0 python test_emu3.py

LlamaGen

CUDA_VISIBLE_DEVICES=0 python test_llamagen.py

Acknowledge

Our code is based on Lumina-mGPT, Emu3, LlamaGen, and Anole. We would like to express our gratitude to Tianwei Xiong for his assistance.