search

ZYM-PKU / UDiffText

UDiffText: A Unified Framework for High-quality Text Synthesis in Arbitrary Images via Character-aware Diffusion Models
MIT License
207 stars 17 forks source link

How long does it take to train this model? #13

Open TongkunGuan opened 4 months ago

TongkunGuan commented 4 months ago

It's a good work! I would like to know how long it will take to train this model.

ZYM-PKU commented 4 months ago

Thanks. Under current implementation, it takes about 48 hrs to train on the LAION-OCR dataset (batchsize=64, iters=100k, gpu=NVIDIA A40)

TongkunGuan commented 4 months ago

Thanks. Under current implementation, it takes about 48 hrs to train on the LAION-OCR dataset (batchsize=64, iters=100k, gpu=NVIDIA A40)

Thanks for your reply！ How many GPUs are needed for 48h training?

ZYM-PKU commented 4 months ago

I used 8*A40