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
I appreciate your awesome work very much. However, there are some things I don't understand.
During the training process, each iteration has a 20% probability that mask_spatial_lora is True, and mask_temporal_lora is always False. If a certain lora is masked, the lora scale of this part will be set to 0. This does not freeze the lora, but cancels the lora. This is inconsistent with the dual channel mentioned in the paper. Is it because I didn’t understand the code?
thank you for your reply
The text was updated successfully, but these errors were encountered:
Hi. This trick is just for faster training. You can simply set the probability to 0%.
Sorry, maybe I didn't make it clear what I meant. I mean when training, if you want to freeze spatial lora, you set lora_i = 0. But I think this operation only bypasses lora, rather than freezing lora.
I appreciate your awesome work very much. However, there are some things I don't understand.
During the training process, each iteration has a 20% probability that mask_spatial_lora is True, and mask_temporal_lora is always False. If a certain lora is masked, the lora scale of this part will be set to 0. This does not freeze the lora, but cancels the lora. This is inconsistent with the dual channel mentioned in the paper. Is it because I didn’t understand the code?
thank you for your reply
The text was updated successfully, but these errors were encountered: