-
Notifications
You must be signed in to change notification settings - Fork 220
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
Offline batch inference mode #82
Conversation
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 Pat. Not sure if this PR was ready or not, but I left some comments.
Sorry for the few iterations. Ended up working out a batch implementation that doesn't use dataloader. This was because it was tricky to use dataloader effectively without introducing awkwardness in the dataprocessing in infer_tools, and function definitions for batch_infer. |
Thanks Patrick, looks good. Just a few minor comments. |
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.
Looking much better after the other reviews!
I added some comments but they are mostly nit, so it's approved IMO.
* Batch working! Next keep track of indicies in dataset * Batching working * Add some print statements * Add yield * Remove print statements * Fixing types * Small cleanup * More small fixes * Pull out additional data before running through collator * remove _process_dataset_batch unused. * Adding displacement to audio token start idx if padding left is true * Address comments * Addressing comments * Batch without dataloader * Clean up * Remove tensor conversion in collator * Remove extra lines * Addressing comments * Address comments * Using dataloader without collator * Small cleanup * Addressing comments * Remove text output from json * Use zip
Adds batching capabilities.
addresses issue: #68