Different resutls for trained model and model loaded from checkpoint #19864
Unanswered
d3mage
asked this question in
code help: CV
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Hi. I am currently working with brain tumor segmentation with a custom dataset. During the training, my model has been outputting good results for training, testing, and validation data. However, when I load the model with the checkpoint, the same testing data from the training step now produces bad results.
I have tried to resolve the issue by experimenting with the backbone, encoders, optimizers, hyperparameters, etc. Nothing worked and, lastly, I have conducted an experiment where I trained a small model and forwarded dummy data expecting to get the same output.
Training loop
Loading model
Model
Why do I get different results? How do I make the results consistent? Thank you in advance.
Beta Was this translation helpful? Give feedback.
All reactions