Skip to content

Classification prompt tuning#144

Closed
Ssukriti wants to merge 72 commits intocaikit:mainfrom
Ssukriti:classification_prompt_tuning
Closed

Classification prompt tuning#144
Ssukriti wants to merge 72 commits intocaikit:mainfrom
Ssukriti:classification_prompt_tuning

Conversation

@Ssukriti
Copy link
Copy Markdown
Collaborator

@Ssukriti Ssukriti commented Aug 24, 2023

For: #133

Save and load is blocked on next caikit release with this fix that will aloow to save base model caikit/caikit#431

I was not able to ad WIP decorator on run as it gave errors on inspecting run signature. I have added it to all other functions.

@Ssukriti Ssukriti force-pushed the classification_prompt_tuning branch from d469e0f to e054981 Compare August 24, 2023 00:44
Ssukriti and others added 25 commits August 23, 2023 18:45
@Ssukriti
classification prompt tuning wrapper
173ca5f 
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@gkumbhat @Ssukriti
♻️ Refactor trainer logic and move it to resources
cbd7b30 
Signed-off-by: gkumbhat <kumbhat.gaurav@gmail.com>
@gkumbhat @Ssukriti
🚧 Work in progress causal-lm trainer
89ae923 
Signed-off-by: gkumbhat <kumbhat.gaurav@gmail.com>
@gkumbhat @Ssukriti
🎨 Fix linting and formatting
947b3ca 
Signed-off-by: gkumbhat <kumbhat.gaurav@gmail.com>
@gkumbhat @Ssukriti
🐛 Fix seq2seq training arguments
034e746 
Signed-off-by: gkumbhat <kumbhat.gaurav@gmail.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@gkumbhat @Ssukriti
🐛 Remove task ids from resource tokenization functions
7450ff8 
Signed-off-by: gkumbhat <kumbhat.gaurav@gmail.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@gkumbhat @Ssukriti
✅ Add cuda device fixture to get around cuda unit testing when available
29b5dfa 
Signed-off-by: gkumbhat <kumbhat.gaurav@gmail.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@gkumbhat @Ssukriti
🎨 Fix formatting
0a6855b 
Signed-off-by: gkumbhat <kumbhat.gaurav@gmail.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@gkumbhat @Ssukriti
✅✨ Make fine-tuning work for causal lm models and make tests pass
4b2d22c 
Signed-off-by: gkumbhat <kumbhat.gaurav@gmail.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@gkumbhat @Ssukriti
🔧 Make review changes and add docstring for arguments
6643e6f 
Signed-off-by: gkumbhat <kumbhat.gaurav@gmail.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@gkumbhat @Ssukriti
✨ Add support for model.generate right after training by saving and l…
867b950 
…oading the model

Signed-off-by: gkumbhat <kumbhat.gaurav@gmail.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@gkumbhat @Ssukriti
🎨 Fix formatting and linting
d21cf69 
Signed-off-by: gkumbhat <kumbhat.gaurav@gmail.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@gkumbhat @Ssukriti
🐛 Fix default verbalizer declaration in fine_tuning module
e366a9d 
Signed-off-by: gkumbhat <kumbhat.gaurav@gmail.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@gkumbhat @Ssukriti
🔧 Update parameters for trainer and add random seed
7f8c3c4 
Signed-off-by: gkumbhat <kumbhat.gaurav@gmail.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@gkumbhat @Ssukriti
⏪ Revert back dump_api script changes
5420850 
Signed-off-by: gkumbhat <kumbhat.gaurav@gmail.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@gkumbhat @alex-jw-brooks @Ssukriti
Apply suggestions from code review
ac5f4a5 
Co-authored-by: Alex Brooks <alex.brooks@ibm.com>
Signed-off-by: Gaurav Kumbhat <kumbhat.gaurav@gmail.com>
Signed-off-by: gkumbhat <kumbhat.gaurav@gmail.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@Ssukriti
text generation local block
fd84e8d 
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@Ssukriti
formatter fixes
544dcb2 
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@Ssukriti
refactoring text geenration blocks
7b49859 
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@gkumbhat @Ssukriti
🚚 Fix interface imports
5ce846c 
Signed-off-by: gkumbhat <kumbhat.gaurav@gmail.com>
@gkumbhat @Ssukriti
🎨 Fix linting
c38b530 
Signed-off-by: gkumbhat <kumbhat.gaurav@gmail.com>
@gkumbhat @Ssukriti
⚰️ Remove unused code
cedd9f8 
Signed-off-by: gkumbhat <kumbhat.gaurav@gmail.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@gkumbhat @Ssukriti
🐛 Remove default for artifact_path from config in text-gen model load
8808c85 
Signed-off-by: gkumbhat <kumbhat.gaurav@gmail.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@rawkintrevo @Ssukriti
Update README.md
84991c4 
Signed-off-by: Trevor Grant <trevor.d.grant@gmail.com>

Signed-off-by: Trevor Grant <trevor.d.grant@gmail.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@rawkintrevo @gkumbhat @Ssukriti
Update README.md
967cdf1 
Co-authored-by: Gaurav Kumbhat <kumbhat.gaurav@gmail.com>
Signed-off-by: Trevor Grant <trevor.d.grant@gmail.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
gabe-l-hart and others added 27 commits August 23, 2023 19:18
@gabe-l-hart @Ssukriti
TgisAutoFinder: Add the ability to give a specific TGIS backend to us…
99abc9f 
…e from the finder

This requires that the module know that it may receive an in-memory
ModuleConfig which may have a tgis_backend attribute set

Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@gabe-l-hart @Ssukriti
TgisAutoFinder: Skip peft training tests on arm processors
b453fe3 
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@gabe-l-hart @Ssukriti
TgisAutoFinder: Happy path unit tests for the auto finder
8a40854 
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@gabe-l-hart @Ssukriti
TgisAutoFinder: Bump caikit to allow 0.15
c62a5f4 
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@gabe-l-hart @Ssukriti
TgisAutoFinder: Don't require instance_name in finder constructor
d75b7a8 
This is there to meet the factory abstract interface, but it's not used so
if someone ever did want to create an instance manually (which is a valid
usecase when using `caikit.load` directly), we don't want to require it.

Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@gabe-l-hart @Ssukriti
TgisAutoFinder: Bump caikit-tgis-backend for connection testing
5d31306 
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@gabe-l-hart @Ssukriti
TgisAutoFinder: Add model_id in mock TGISConnection
c04ec97 
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@gabe-l-hart @Ssukriti
TgisAutoFinder: Handle bad tgis connections
a8060d0 
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@gabe-l-hart @Ssukriti
TgisAutoFinder: Linting fixes
58a0839 
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>

Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@olson-ibm @Ssukriti
fixed simple typo (caikit#128)
57717cc 
Fix typo in run fine tuning

Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@gkumbhat @Ssukriti
🧑‍💻 Update fine-tuning example to allow local save of the model and f…
c9ff043 
…lexible target seq len

Signed-off-by: gkumbhat <kumbhat.gaurav@gmail.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@gkumbhat @Ssukriti
🎨 fix formatting
4f6bc28 
Signed-off-by: gkumbhat <kumbhat.gaurav@gmail.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@dependabot @Ssukriti
⬆️ Update caikit[runtime-grpc,runtime-http] requirement
7842323 
Updates the requirements on [caikit[runtime-grpc,runtime-http]](https://github.com/caikit/caikit) to permit the latest version.
- [Release notes](https://github.com/caikit/caikit/releases)
- [Changelog](https://github.com/caikit/caikit/blob/main/releases.md)
- [Commits](caikit/caikit@v0.15.0...v0.16.0)

---
updated-dependencies:
- dependency-name: caikit[runtime-grpc,runtime-http]
  dependency-type: direct:production
...

Signed-off-by: dependabot[bot] <support@github.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@tharapalanivel @Ssukriti
⬆️ upgrade caikit
1c05d26 
Signed-off-by: Thara Palanivel <130496890+tharapalanivel@users.noreply.github.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@tharapalanivel @Ssukriti
🚧 Rebase on caikit interfaces
6ed7486 
Signed-off-by: Thara Palanivel <130496890+tharapalanivel@users.noreply.github.com>
@tharapalanivel @Ssukriti
⬆️ Bump caikit-tgis-backend
fe732fc 
Signed-off-by: Thara Palanivel <130496890+tharapalanivel@users.noreply.github.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@gkumbhat @Ssukriti
Update CODEOWNERS
b432a07 
Signed-off-by: Gaurav Kumbhat <kumbhat.gaurav@gmail.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@Ssukriti
add unit tests
ecbc749
@Ssukriti
load and save tests
739cc61 
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@Ssukriti
classification prompt tuning block
e2868ec 
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@jolson-ibm @Ssukriti
Closes caikit#134
458cee9 
Signed-off-by: Joe Olson <joe.olson@ibm.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@jolson-ibm @Ssukriti
Closes caikit#134
02ce643 
Signed-off-by: Joe Olson <joe.olson@ibm.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@jolson-ibm @Ssukriti
Closes caikit#134
d6a8f94 
Signed-off-by: Joe Olson <joe.olson@ibm.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@gkumbhat @Ssukriti
🚧 Debug streaming API generation
7900bf0 
Signed-off-by: gkumbhat <kumbhat.gaurav@gmail.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@gkumbhat @Ssukriti
📦 Update caikit
c574b13 
Signed-off-by: gkumbhat <kumbhat.gaurav@gmail.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@gkumbhat @Ssukriti
📝 Add comment explaining wip comment for prompt tuning
a1022d3 
Signed-off-by: gkumbhat <kumbhat.gaurav@gmail.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@dependabot @Ssukriti
⬆️ Update caikit[runtime-grpc,runtime-http] requirement
e732057 
Updates the requirements on [caikit[runtime-grpc,runtime-http]](https://github.com/caikit/caikit) to permit the latest version.
- [Release notes](https://github.com/caikit/caikit/releases)
- [Changelog](https://github.com/caikit/caikit/blob/main/releases.md)
- [Commits](caikit/caikit@v0.16.0...v0.17.0)

---
updated-dependencies:
- dependency-name: caikit[runtime-grpc,runtime-http]
  dependency-type: direct:production
...

Signed-off-by: dependabot[bot] <support@github.com>
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
@Ssukriti Ssukriti force-pushed the classification_prompt_tuning branch from e054981 to e732057 Compare August 24, 2023 01:30
@Ssukriti
Copy link
Copy Markdown
Collaborator Author

Ssukriti commented Aug 24, 2023

something went wrong in rebase, closing in favor of #145

@Ssukriti Ssukriti closed this Aug 24, 2023
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@alex-jw-brooks alex-jw-brooks Awaiting requested review from alex-jw-brooks

@gkumbhat gkumbhat Awaiting requested review from gkumbhat gkumbhat is a code owner

@evaline-ju evaline-ju Awaiting requested review from evaline-ju evaline-ju is a code owner

@gabe-l-hart gabe-l-hart Awaiting requested review from gabe-l-hart gabe-l-hart is a code owner

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

7 participants

@Ssukriti @gkumbhat @rawkintrevo @gabe-l-hart @olson-ibm @tharapalanivel @jolson-ibm