Query about trainable parameters initialization

Question

Query about trainable parameters initialization

Closed this issue a year ago · 3 comments

Thanks for your interesting work. As mentioned in the experimental settings of the paper, the trainable prompt learning parameters are initialized with xavier uniform initialization scheme. Where is the code that initializes the trainable parameters?
If you can reply my queries, I would appreciate it very much.

Answer 1 · 2023-09-16T02:27:55.000Z

Thanks for your interesting work. As mentioned in the experimental settings of the paper, the trainable prompt learning parameters are initialized with xavier uniform initialization scheme. Where is the code that initializes the trainable parameters? If you can reply my queries, I would appreciate it very much.

Did you solve this problem, I had the same problem as you, I didn't find the setting in the code for the trainable parameter. Looking forward to hearing from you, thanks.

Answer 2 · 2023-09-16T04:23:27.000Z

The function get_optimizer_scheduler located in lib/train/base_functions.py determines which parameters are trainable. Hope this helps.

Answer 3 · 2023-09-16T07:04:43.000Z

The function get_optimizer_scheduler located in lib/train/base_functions.py determines which parameters are trainable. Hope this helps.

It helps! Thanks for your reply. I hope you have a wonderful day!