search

plai-group / flexible-video-diffusion-modeling

MIT License
111 stars 14 forks source link

Total training time #8

Closed tarashakhurana closed 4 months ago

tarashakhurana commented 9 months ago

Hi, for the experiments reported in your paper, how much total training time does your model take?

wenyufeng936 commented 4 months ago

Hello, how much time did you spend training the model when reproducing this paper?

wsgharvey commented 4 months ago

Hi, sorry for the slow response. This is in Appendix A of https://arxiv.org/pdf/2205.11495 (we report GPU-hours and number of GPUs). Our largest scale experiment on CARLA Town01 used 380 GPU-hours parallelised on 4 GPUs, so 95 hours.