-
Notifications
You must be signed in to change notification settings - Fork 35
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
added tokenization tasks to serve tokenization requests #325
Conversation
Signed-off-by: Shonda-Adena-Witherspoon <shonda.adena.witherspoon@ibm.com>
6c91a90
to
dc8cbd3
Compare
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 for the contribution! A couple asks:
- A few small unit tests would be helpful for the new module function. We should strive for tests on any new functionality. A section like for the streaming tests https://github.com/caikit/caikit-nlp/blob/main/tests/modules/text_generation/test_text_generation_tgis.py#L135-L161 on each module might be useful to reference.
- We will likely want to allow for the tokenization to also run on prompt tuned models so it'd be good to have the same update here: https://github.com/caikit/caikit-nlp/blob/main/caikit_nlp/modules/text_generation/peft_tgis_remote.py
- [After all updates] For releasing we will want this duplicated to the
release-0.3
branch in order to also release this functionality without embeddings.
…task in decorator Signed-off-by: Shonda-Adena-Witherspoon <shonda.adena.witherspoon@ibm.com>
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 Shonda, I did an initial review and added a few comments. Can you please add unit tests around tokenization for the PeftPromptTuningTGIS
module too?
…e.py Signed-off-by: Shonda-Adena-Witherspoon <shonda.adena.witherspoon@ibm.com>
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.
one tiny nit but otherwise lgtm! You'll need a duplicate PR to release-0.3
as well for now
Signed-off-by: Shonda-Adena-Witherspoon <shonda.adena.witherspoon@ibm.com>
@tharapalanivel should I make the duplicate PR for the |
@evaline-ju - merging is still block despite Thara's approval for the changes you requested. Not sure how to proceed |
Added tokenization tasks in order to serve tokenization request and return the token count