You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Dear authors,
You claimed in the paper that "using a single NVIDIA RTX-3090 GPU with a training batch size of 4, employing learning rates of 4 × 10−6 for DB", is that correct?
I found it slow on my V100GPU under this hyper parameter setting(two and an half hour for training a single concept),my gradient accumulation steps is set to 1, and max_train_steps is 1500, I would appreciate it if you could help me
The text was updated successfully, but these errors were encountered:
Dear authors,
You claimed in the paper that "using a single NVIDIA RTX-3090 GPU with a training batch size of 4, employing learning rates of 4 × 10−6 for DB", is that correct?
I found it slow on my V100GPU under this hyper parameter setting(two and an half hour for training a single concept),my gradient accumulation steps is set to 1, and max_train_steps is 1500, I would appreciate it if you could help me
The text was updated successfully, but these errors were encountered: