🐛 Bug Trainer() argument name ch

Thanks for reporting this ! Are those deprecated already? Or <code class="notransl

Closed by <a class="issue-link js-issue-link" data-error-text="Failed to load title" d

pytorch-lightning deprecations about asteroid HOT 5 CLOSED

asteroid-team commented on July 3, 2024

pytorch-lightning deprecations

from asteroid.

Comments (5)

mpariente commented on July 3, 2024

Thanks for reporting this !
Are those deprecated already? Or DeprecationWarning is thrown?

I intended to upgrade to 0.8.1 because they have new features that will be helpful.
I'll do this pretty soon.

from asteroid.

jonashaag commented on July 3, 2024

They are removed in 0.8

With this patch training works for me so far

diff --git a/egs/wham/DPRNN/train.py b/egs/wham/DPRNN/train.py
index b4ebeb8..6b7020f 100644
--- a/egs/wham/DPRNN/train.py
+++ b/egs/wham/DPRNN/train.py
@@ -75,10 +75,10 @@ def main(conf):

     # Don't ask GPU if they are not available.
     gpus = -1 if torch.cuda.is_available() else None
-    trainer = pl.Trainer(max_nb_epochs=conf['training']['epochs'],
+    trainer = pl.Trainer(max_epochs=conf['training']['epochs'],
                          checkpoint_callback=checkpoint,
                          early_stop_callback=early_stopping,
-                         default_save_path=exp_dir,
+                         default_root_dir=exp_dir,
                          gpus=gpus,
                          distributed_backend='ddp',
                          gradient_clip_val=conf['training']["gradient_clipping"])

from asteroid.

mpariente commented on July 3, 2024

Ok thanks !

from asteroid.

mpariente commented on July 3, 2024

I let you make a PR in setup.py and requirements?
For now, let's pin it

from asteroid.

mpariente commented on July 3, 2024

Closed by #159
Thanks Jonas

from asteroid.

Recommend Projects