v0.2.1 release

V1beta3 (google-gemini#67)

* Working on v1beta3

Add support for batched embeddings
Update to v0.2
Add get_tuned_model, get_base_model, get_model.
Add list_models, list_tuned_models.
Add create_tuned_model.
Support batched embeddings.

Change-Id: I917576a61719fcd90cca581506e15826b67cae88

* Implement {update,delete}_tuned_model.

Change-Id: Idd276934666f7fa8da034e59a80e4e6047cea4d5

* Decode TunedModel to python objects.

Change-Id: I4376012686b53065f9372bd8dd6a340d3e0ca43a

* Add more model service tests + bug fixes + pytype.

Change-Id: Iaa1b4e75b5c075d1b0b6b383d37da8d254b0715b

* Upgrade the Operation returned by create_tuned_model.

Updated into colab here: https://colab.sandbox.google.com/drive/1Eros5eLfAl4pNfsQ2sftyx5khq2TrK_V?resourcekey=0-ByOp_F6lxWCuKVJM_tUjeQ

Change-Id: Ic3e40f25abc6eba69a8866fe855c7ecb83c4c4c3

* Implement the operations service.

Change-Id: I29e92a4c8697f165deb9782599b21d98eb869737

* fix update_tuned_model

Change-Id: Iaacaaa08c1807e13ad520804dd20eb1e54a9b34d

* Remove decoding error workaround + format + use the 0.3.2 v1beta3 package.

Change-Id: I6e329c635aa51e67c50b2318a506455c21c8d988

* Fix tests & remove pytz

Fix tuned_model/base_model/source_model handling so nested tuned models work
Remove the last 'v1main'

Change-Id: Ia07c169d0bfb0782edd0115b5d0418c9eb3eb040

* black .

* fix test

* add tqdm requirement

* Use union to fix py3.9

* Use union to fix py3.9

* black .

Check format with black. (google-gemini#31)

* Add a GH action to run the tests.

* Add a action to check format with black.

* format with black

---------

Co-authored-by: Mark McDonald <[email protected]>

Update version.py (google-gemini#24)

Test chat citations.

First commit: Push everyone's code.

Co-authored-by: Mark McDonald <[email protected]>
Co-authored-by: Shilpa Kancharla <[email protected]>
Co-authored-by: Petchean Ang <[email protected]>