denoising-diffusion-pytorch icon indicating copy to clipboard operation
denoising-diffusion-pytorch copied to clipboard

Published 20 hours ago • verified publisher icon lucidrains

problem

Open HeliaHashemipour opened this issue 11 months ago • 1 comments

Hi, Can you tell me what is train_num_steps and timesteps? I can't get the difference between them. (which one of them is the timesteps in the paper?)

HeliaHashemipour avatar Mar 10 '24 10:03 HeliaHashemipour

train_num_steps is the number of times the gradient is computed (for every train_batch_size * gradient_accumulate_every samples in your dataset) until the model stops training

timesteps is "the one from the paper"

dancingacidpanda avatar Mar 14 '24 21:03 dancingacidpanda