Why is the output of the model NaN when running train.py?

[AlexZzy1985]

When i run the follwoing commant: python lmpt/train.py
--dataset 'voc-lt'
--seed '0'
--pretrain_clip 'ViT16'
--batch_size 64
--epochs 50
--class_token_position 'end'
--ctx_init ''
--n_ctx 16
--m_ctx 2
--training_method 'lmpt'
--lr 5e-4
--loss_function dbl
--cseloss softmargin
--optimizer sgd
--neg_scale 2.0
--gamma 0.2
--lam 0.5
the output of 'CustomCLIP' is NaN. I've carefully checked the code and found that in CustomCLIP, there are NaNs in the result of prompts = self.prompt_learner(). I'm wondering if there are any friends who have encountered similar problems. I'm really looking forward to getting your help. Also, thanks to the author for the outstanding work!