I’m trying to create callback that will use current state of model when training ends. I’m using “on_fit_end” method and calling model(data) to create prediction. Should I also switch model mode to “eval()” or model is in that state on this stage of training?
Related topics
Topic | Replies | Views | Activity | |
---|---|---|---|---|
Are on_fit_end and on_train_end the same? | 4 | 5509 | October 19, 2023 | |
Reloading model with trainer.fit(ckpt_path) and overrides callback | 0 | 316 | August 14, 2023 | |
How to call the validation and ModelCheckpoint callback after certain epochs?
|
0 | 148 | September 14, 2023 | |
Call optimizer_step inside train_epoch_end()
|
0 | 374 | August 9, 2021 | |
Retraining a model with new data | 1 | 389 | May 9, 2023 |