qrzou / parco Goto Github PK

Hello,can you tell me what gpu do you use when you train Parco_trans and how long it takes?When I train the second part using RTX3090 , I meet the question that cuda is out of memory.Thank you!

Hi,

Thanks for this wonderful code, may I ask how can I train the stage 2 - Parco_trans on 2 GPUs?

many thanks!

Thank you for your great works!

I have one question.
As I run your code in this repository, I see that the performance was as good as described in the paper.

You used the code below to control the randomness.
'''
torch.manual_seed(args.seed)
'''
But I know that it is difficult to control randomness with this alone.
In reality, the performance was slightly different each time I trained.

So, I insert below code to control the randomness.
'''
random.seed(seed)
np.random.seed(seed)
torch.manual_seed(seed)
torch.cuda.manual_seed_all(seed)
torch.backends.cudnn.deterministic = True
torch.backends.cudnn.benchmark = False
'''
However, by doing this, the learning time slowed down by more than 100 times in vq training.

If you know the cause of this, I would appreciate it if you could explain.

Thank you!

Hi there,

Thanks for sharing the code, actually I found that the repository only contains the KIT training details, would you mind sharing the HumanML3D setting?

Very awesome idea！I am very interested about your body partition. Where can I find your supplementary materials？Thank you for your reply very much!!!