gpus

[rohitsaluja22]

Hi may I know which gpus are used, along with numbers, for training in following papers?

1. Bao, Wentao, Qi Yu, and Yu Kong. "Uncertainty-based traffic accident anticipation with spatio-temporal relational learning." Proceedings of the 28th ACM International Conference on Multimedia. 2020.
2. Bao, W., Yu, Q. and Kong, Y., 2021. DRIVE: Deep reinforced accident anticipation with visual explanation. In Proceedings of the IEEE/CVF International Conference on Computer Vision (pp. 7619-7628).

Thanks

[Cogito2012]

@rohitsaluja22 I was using Titan RTX (or Titan Xp) for training the UString model, and RTX 3090 and A100 for training the DRIVE model.
I simply use a single GPU in each training task.

[rohitsaluja22]

@Cogito2012 Thanks for your reply
You mean 1 Tensor RTX (24 GB)/Titan XP (12 GB) for UString. Will 12 GB RAM be enough?
Also, RTX 3090 has 24 GB RAM, and A100 you used was 32 GB or 64 GB? Please clarify you mean RTX 3090 and A100? Or RTX 3090 or A100!

[Cogito2012]

@rohitsaluja22 12GB for UString and 24G for the DRIVE should be enough. For the A100, I used a 32 GB version. Each time I train the model, I use one GPU so either one RTX 3090 or one A100 is enough.

[mwy001]

@Cogito2012 How much time does it cost to train the UString model on crash dataset with your hardware setting?
(command: "bash run_train_test.sh train 0 crash 10" with default epoch=30)