Replies: 1 comment
-
Sorry, I looked at the train code and realized that this is only true if freeze epoch is greater than 0. |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
-
Hi, I have a question regarding the training stages and when the image encoder is actually trained.
Looking at the train_config for stages 1 to 4:
Up to stage 2, the auto component is frozen.
Up to stage 4, the point component is frozen.
However, based on the internal code, it seems that if either auto or point is frozen, the image encoder is also frozen.
Given this, I'm not sure when exactly the image encoder gets updated (i.e., trained) during the training process. Could you clarify this?
Thanks in advance!
Beta Was this translation helpful? Give feedback.
All reactions