-
Notifications
You must be signed in to change notification settings - Fork 427
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Two step source separation #67
Two step source separation #67
Conversation
…d deviation as an oprion.
…rating the two step source separation recipe. Specifically, for the two step process a new method callback is called which has a modular structure and does not intervene with the other callback methods existing.
…opriate format and docs. Moreover we add a simpler block for dilated separable convolutions without skip connections.
…zation for the filterbank and the separator.
…ses for loading the appropriate module parts and encoding directories paths.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks a lot for the PR, it will be a great addition to the set of recipes we have !
There are few things to address before merging but it looks great already 🚀
…d deviation as an oprion.
…rating the two step source separation recipe. Specifically, for the two step process a new method callback is called which has a modular structure and does not intervene with the other callback methods existing.
…opriate format and docs. Moreover we add a simpler block for dilated separable convolutions without skip connections.
…zation for the filterbank and the separator.
…ses for loading the appropriate module parts and encoding directories paths.
…d deviation as an oprion.
…opriate format and docs. Moreover we add a simpler block for dilated separable convolutions without skip connections.
…zation for the filterbank and the separator.
…ses for loading the appropriate module parts and encoding directories paths.
…rating the two step source separation recipe. Specifically, for the two step process a new method callback is called which has a modular structure and does not intervene with the other callback methods existing.
… error as requested
…problem with a background running server job
…ive encoder/decoder
…ore code as specified. However, in a next PR we can move the two-step tdcn inside the main block codebase as it is almost ready
@@ -52,24 +51,29 @@ def forward(self, *args, **kwargs): | |||
""" | |||
return self.model(*args, **kwargs) | |||
|
|||
def common_step(self, batch, batch_nb): | |||
def common_step(self, batch, batch_nb, train=True): |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I guess you cherry-picked three of my commits to have the new System
.
This is bad habit for two reasons
- The changes will be asociated with this PR (wrong history)
- If you make a small change on this new version of the file, it is not as obvious to see
There are two ways to deal with this - Rebase master and force-push
- Merge master into your branch
Because we squash PRs before merging, I prefer the second method.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Oh I am sorry man this was a mistake
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
That looks good, thanks ! We'll keep it in the recipe for now and try to merge the architecture definitions to avoid redundancy.
Thanks again !
Integrating the two-step source separation in asteroid.