Commit Graph
3871 Commits
Author SHA1 Message Date
22709379dd feat: Add prediction dedicated endpoint colab sample (#3942)
* feat: Add prediction dedicated endpoint colab sample

* Update get_started_with_dedicated_endpoint.ipynb

---------

Co-authored-by: TJ(Tianjiao) Liu <tianjiaoliu@google.com>
2025-05-15 01:14:40 +00:00
Vertex MG TeamandCopybara-Service 4aafcfb40f Add qwen3 specific instruction to runtime creation.
PiperOrigin-RevId: 758737197
2025-05-14 10:21:56 -07:00
Rayan DasoriyaandCopybara-Service d53aa0c816 Fix gemma2 notebook
PiperOrigin-RevId: 758283299
2025-05-13 10:43:57 -07:00
Rayan DasoriyaandCopybara-Service 21976bf94f Clarify the HF token requirement
PiperOrigin-RevId: 758277776
2025-05-13 10:33:15 -07:00
Vertex MG TeamandCopybara-Service 434e8ac8fc [Fix] Add NVIDIA_A100_80GB option to vertex finetuning.
PiperOrigin-RevId: 757854822
2025-05-12 11:55:28 -07:00
Vertex MG TeamandCopybara-Service 2376532e3d Fix A100_80GB machine type.
PiperOrigin-RevId: 757734855
2025-05-12 06:17:50 -07:00
Vertex MG TeamandCopybara-Service 8b2fbe3f34 Make runtime connection instructions clearer for axolotl notebook.
PiperOrigin-RevId: 757720267
2025-05-12 05:29:51 -07:00
Vertex MG TeamandCopybara-Service f991ae44ab Fix local finetuning for axolotl.
PiperOrigin-RevId: 757655772
2025-05-12 02:01:51 -07:00
Dustin LuongandCopybara-Service 7eb8b76d19 Remove reference server deployment and inference from llama3.2 notebook.
PiperOrigin-RevId: 757455670
2025-05-11 11:01:47 -07:00
Vertex MG TeamandCopybara-Service 7da9c0b644 Remove a3-ultragpu from Axolotl notebook.
PiperOrigin-RevId: 756910136
2025-05-09 14:35:55 -07:00
Genquan DuanandCopybara-Service 46a75498d2 Update notebook to add more machine type suggestions, and support allowlisted real-time forecasting for graph_operational model.
PiperOrigin-RevId: 756860759
2025-05-09 12:17:35 -07:00
Vertex MG TeamandCopybara-Service c2208eb454 Fix hf cache dir for local finetuning and use gcsfuse for local training and merging of model.
PiperOrigin-RevId: 756772796
2025-05-09 08:15:32 -07:00
Vertex MG TeamandCopybara-Service f9c43d4a8a Add a3-ultragpu with dws to Axolotl notebook.
PiperOrigin-RevId: 756625405
2025-05-08 23:39:48 -07:00
Vertex MG TeamandCopybara-Service 32ae5b4af0 Add gpu type check in axolotl local run.
PiperOrigin-RevId: 756598801
2025-05-08 22:04:50 -07:00
Vertex MG TeamandCopybara-Service 2345f89625 Refactoring and minor fixes
PiperOrigin-RevId: 756570419
2025-05-08 20:25:13 -07:00
Vertex MG TeamandCopybara-Service 46a80b731b Add Dia-1.6B serving notebook
PiperOrigin-RevId: 756499377
2025-05-08 16:13:17 -07:00
Vertex MG TeamandCopybara-Service 2781808a96 Add n1+T4 as one deployment option for e5 model.
PiperOrigin-RevId: 756479268
2025-05-08 15:16:27 -07:00
Vertex MG TeamandCopybara-Service 5af1bc523d Fix Axolotl installation.
PiperOrigin-RevId: 756401696
2025-05-08 11:54:00 -07:00
Rayan DasoriyaandCopybara-Service 9742d29e51 Add eval harness support for Qwen2.5 finetuning notebook
PiperOrigin-RevId: 756350741
2025-05-08 09:50:30 -07:00
Vertex MG TeamandCopybara-Service 30fda48397 change default gpu type as empty to avoid unintended gpu runtime creation
PiperOrigin-RevId: 756199290
2025-05-08 01:17:09 -07:00
Vertex MG TeamandCopybara-Service 6888728be5 Fix dataset vars definition issue and make runtime creation session optional in Axolotl notebook.
PiperOrigin-RevId: 756138907
2025-05-07 21:40:57 -07:00
Vertex MG TeamandCopybara-Service 8c7fbc6210 Phi-4 reasoning variants
PiperOrigin-RevId: 756047269
2025-05-07 16:17:35 -07:00
Dustin LuongandCopybara-Service 4c4e224d31 Update Qwen3 deployment notebook with FP8 support.
PiperOrigin-RevId: 756041267
2025-05-07 15:59:11 -07:00
Vertex MG TeamandCopybara-Service ac98f72005 Fix runtime creation codes in Axolotl notebook to disallow L4x16 combination.
PiperOrigin-RevId: 755968781
2025-05-07 12:51:45 -07:00
Vertex MG TeamandCopybara-Service ddebceb70c Support Workbench in Gemma 2 tuning notebook
PiperOrigin-RevId: 755884581
2025-05-07 09:24:14 -07:00
Vertex MG TeamandCopybara-Service d8e5c3b461 Update Axolotl finetuning notebook to support Qwen3.
PiperOrigin-RevId: 755577906
2025-05-06 16:52:10 -07:00
Eric DongandGitHub 0bfe70afcd refactor: remove runtime reboot (#4026)
* refactor: remove runtime reboot

* Remove spaces

* Use %pip instead

* Update model path

* Update BQ path

* Downgrade numpy for backfoward compability
2025-05-05 21:12:55 +00:00
Ravi DalalandGitHub 7488a5dc27 upgraded spark on ray on vertex ai notebook to 2.42.0 version (#4027)
* upgraded spark on ray on vertex ai notebook to 2.42.0 version

* upgraded spark on ray on vertex ai notebook to 2.42.0 version

* upgraded spark on ray on vertex ai notebook to 2.42.0 version

* upgraded spark on ray on vertex ai notebook to 2.42.0 version
2025-05-05 19:05:30 +00:00
Vertex MG TeamandCopybara-Service 4914987998 Update Axolotl docker version for the finetuning notebook
PiperOrigin-RevId: 754984828
2025-05-05 10:20:32 -07:00
Vertex MG TeamandCopybara-Service abd30065d0 Add support for dedicated endpoint in paligemma finetuning
PiperOrigin-RevId: 753860258
2025-05-01 22:26:25 -07:00
Dustin LuongandCopybara-Service 68960f1221 Add SDK deploy option to Qwen3 deployment notebook.
PiperOrigin-RevId: 753636431
2025-05-01 10:17:08 -07:00
Vertex MG TeamandCopybara-Service 836d1ebbd1 No public description
PiperOrigin-RevId: 753635866
2025-05-01 10:15:36 -07:00
Vertex MG TeamandCopybara-Service f96c830a66 chore: Add back faster_deploy_enabled in deploy SDK.
PiperOrigin-RevId: 753454732
2025-04-30 23:17:36 -07:00
Dustin LuongandCopybara-Service ecd1fc28d2 Set model id to GCS bucket for Qwen3 235B model for better stability.
PiperOrigin-RevId: 753430192
2025-04-30 21:33:20 -07:00
Vertex MG TeamandCopybara-Service da0874682f Set dedicate endpoint as default for model_garden_finetuning_tutorial
PiperOrigin-RevId: 753294164
2025-04-30 13:38:52 -07:00
Dustin LuongandCopybara-Service e851c1ad99 Upload Qwen3 deployment notebook.
PiperOrigin-RevId: 753265446
2025-04-30 12:19:21 -07:00
Vertex MG TeamandCopybara-Service 93fd7088ba Add deployment options for the smaller Qwen 2.5 model versions too in the finetuning notebook.
PiperOrigin-RevId: 753246638
2025-04-30 11:30:02 -07:00
Vertex MG TeamandCopybara-Service cab440a05b Fixes for Llama Prompt Guard variant to deployment notebook.
PiperOrigin-RevId: 752852958
2025-04-29 13:08:40 -07:00
Vertex MG TeamandCopybara-Service f03f88a0f7 Add Llama Prompt Guard variant to deployment notebook.
PiperOrigin-RevId: 752802325
2025-04-29 10:58:25 -07:00
Vertex MG TeamandCopybara-Service 841fbf9f53 Add Llama Guard variant to deployment notebook.
PiperOrigin-RevId: 752801901
2025-04-29 10:56:47 -07:00
Vertex MG TeamandCopybara-Service bed0c09ad9 Add Model Garden deploy SDK
PiperOrigin-RevId: 752589085
2025-04-28 22:43:24 -07:00
Vertex MG TeamandCopybara-Service 2497476009 Remove 0.5B and 1.5B Qwen 2.5 models from the notebook
PiperOrigin-RevId: 752453505
2025-04-28 14:42:09 -07:00
Eric DongandGitHub 5e2841384f refactor: Use raw string in regex (#4016)
* refactor: Use raw string in regex

* np.NaN was removed in the NumPy 2.0

* Move default python version to 3.10

* Move back default python version to 3.9

* Remove version mistmatched tes notebooks
2025-04-28 19:54:44 +00:00
49de587b2c Add optional support for dedicated endpoint (#4014)
Co-authored-by: Rayan Dasoriya <dasoriya@google.com>
2025-04-28 12:03:33 +00:00
Vertex MG TeamandCopybara-Service feea47a206 Enable the use of dedicated Endpoints in instructpix2pix notebook
PiperOrigin-RevId: 751641417
2025-04-25 19:45:08 -07:00
Vertex MG TeamandCopybara-Service f4b1b277cf Add HiDream-I1 serving notebook
PiperOrigin-RevId: 751521086
2025-04-25 12:38:34 -07:00
Vertex MG TeamandCopybara-Service e4608c983b chore: Clean up endpoint resource names.
PiperOrigin-RevId: 751415593
2025-04-25 07:58:12 -07:00
Vertex MG TeamandCopybara-Service f147b50332 Support VPC-SC.
PiperOrigin-RevId: 751028021
2025-04-24 09:45:20 -07:00
Changyu ZhuandCopybara-Service b1b16f718b Fix missing positional argument in DeepSeek deployment notebook
PiperOrigin-RevId: 750742323
2025-04-23 15:19:37 -07:00
Vertex MG TeamandCopybara-Service 871eb25dc3 Deprecate the colab notebooks for a few models.
PiperOrigin-RevId: 750692428
2025-04-23 12:54:53 -07:00