Skip to content

bug in folds; time-efficient voting strategy; variable number of channels and prototypes - #11

Open
jgciudad wants to merge 7 commits into
guidogagl:mainfrom
jgciudad:prototype_features
Open

bug in folds; time-efficient voting strategy; variable number of channels and prototypes#11
jgciudad wants to merge 7 commits into
guidogagl:mainfrom
jgciudad:prototype_features

Conversation

@jgciudad

Copy link
Copy Markdown
Contributor

No description provided.

@jgciudad
jgciudad marked this pull request as draft August 20, 2025 14:08
@jgciudad
jgciudad force-pushed the prototype_features branch from a363cec to 4226ac6 Compare August 20, 2025 17:43
@jgciudad
jgciudad marked this pull request as ready for review August 20, 2025 17:46
Comment thread physioex/data/dataset.py

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

When no specific fold is specified (fold=-1), a fold is sampled when the training dataset is created, and again when eval_dataset is created, leading to different folds in each.

I’ve changed the default fold to fold 0 when no fold is specified. I’ve done this everywhere in the code and removed random fold selection.

except :
devices = "auto"
effective_batch_size = batch_size * num_nodes
num_steps = datamodule.__len__() // effective_batch_size

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

effective_batch_size was not defined. not sure if the block just before was removed by mistake, or if effective_batch_size should just be changed to batch_size

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant