how long to train the model. If this code can run on 3090

lhoyer / HRDA

[ECCV22] Official Implementation of HRDA: Context-Aware High-Resolution Domain-Adaptive Semantic Segmentation

Other

231 stars 32 forks source link

how long to train the model. If this code can run on 3090 #14

Closed yuheyuan closed 1 year ago

yuheyuan commented 1 year ago

I am interested in this paper and I knwo transform framwork needs high equipment. So I want to know the time to train it. My device is not good. So I want to ask.

lhoyer commented 1 year ago

You can find the runtime details in Section G in the appendix on arXiv. Training DAFormer takes 16 hours on a single Nivida RTX 2080 Ti. As you have a more powerful RTX 3090, it should be faster on your device.

lhoyer commented 1 year ago

Training HRDA takes 32 hours on a single Nvidia Titan RTX and requires 22.5 GB GPU memory. Further details can be found in Tab. 5 and Sec. F.