Support launching graph store jobs from trainer and inferencer #390
Conversation
|
/unit_test |
|
/integration_test |
GiGL Automation@ 19:43:39UTC : 🔄 @ 21:04:31UTC : ✅ Workflow completed successfully. |
GiGL Automation@ 19:43:50UTC : 🔄 @ 20:50:33UTC : ✅ Workflow completed successfully. |
kmontemayor2-sc
changed the title
There was a problem hiding this comment.
The hope for this file in gigl/common was to operate at a layer above business logic gigl/src/*.
i.e. it can plug and play anywhere, and not reliant / specific to GiGL.
Let's continue to adopt seperation of concerns here i.e. AppliedTaskIdentifier, knowledge of GiGL components, and resource config should not live inside vertex_ai service, and the callers should have logic on how to compose the objects needed to initialize the relevant vertex ai jobs - like they do when calling existing methods for the vertex ai service.
|
Sure, updated to not rely on I know we talked about creating some new file offline but I think creating some new |
python/tests/unit/src/common/translators/vertex_ai_job_translator_test.py
Outdated
Show resolved
Hide resolved
python/tests/unit/src/common/translators/vertex_ai_job_translator_test.py
Outdated
Show resolved
Hide resolved
mkolodner-sc
left a comment
mkolodner-sc
left a comment
There was a problem hiding this comment.
Thanks Kyle! Did a quicker pass here since this has already been approved, a few small comments but generally LGTM once those are addressed/answered
|
/unit_test |
|
/e2e_test |
GiGL Automation@ 22:31:48UTC : 🔄 @ 23:36:07UTC : ✅ Workflow completed successfully. |
GiGL Automation@ 22:31:50UTC : 🔄 @ 22:39:40UTC : ✅ Workflow completed successfully. |
GiGL Automation@ 22:31:54UTC : 🔄 @ 23:52:12UTC : ✅ Workflow completed successfully. |
|
/unit_test |
|
/e2e_test |
GiGL Automation@ 22:40:20UTC : 🔄 @ 23:41:22UTC : ✅ Workflow completed successfully. |
GiGL Automation@ 22:40:24UTC : 🔄 @ 22:50:34UTC : ✅ Workflow completed successfully. |
GiGL Automation@ 22:40:28UTC : 🔄 @ 23:55:15UTC : ✅ Workflow completed successfully. |
mkolodner-sc
left a comment
mkolodner-sc
left a comment
There was a problem hiding this comment.
Thanks a lot Kyle!
kmontemayor2-sc
requested review from
nshah-sc,
xgao4-sc,
yliu2-sc and
zfan3-sc
as code owners
Scope of work done
Support launching graph store jobs from trainer and inferencer
Also break out some shared code into utils :)
No tests since we'll have some bigger e2e tests later.
Where is the documentation for this feature?: N/A
Did you add automated tests or write a test plan?
Updated Changelog.md? NO
Ready for code review?: NO