Fix gpu integ test failure due to outdated MPG

[rsareddy0329]

Issue #, if available:

Description of changes:

• test *

======================================================================= test session starts =======================================================================
platform darwin -- Python 3.12.10, pytest-8.4.2, pluggy-1.6.0 -- /Users/rsareddy/workplace/virtual_envs/sagemaker-v3/bin/python3.12
cachedir: .pytest_cache
rootdir: /Users/rsareddy/workplace/sagemaker-sdk-forked-mtrl-vpc-support/sagemaker-python-sdk/sagemaker-train
configfile: pyproject.toml
plugins: cov-7.0.0, anyio-4.13.0
collected 1 item                                                                                                                                                  

tests/integ/train/test_mtrl_trainer_integration.py::TestMTRLEvalIntegration::test_evaluate_finetuned_model PASSED                                           [100%]

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

[github-actions]

Claude finished @rsareddy0329's task in 22s —— View job

---

PR Review: Fix gpu integ test failure due to outdated MPG

• Read PR diff
• Review the change for correctness & conventions

Summary

The change updates the PROD account (729646638167) model_package_group ARN in the integ test config from openai-reasoning-gpt-oss-20b-mtrl-mpg to mock-oss-test-mtrl-mpg.

This looks correct and is consistent with the surrounding PROD config, which already uses the mock-oss-test naming for base_model ("mock-oss-test") and existing_job_name ("mock-oss-test-mtrl-..."). The old MPG name referenced openai-reasoning-gpt-oss-20b, which no longer matches the model actually used in that account — so this alignment fixes the outdated reference.

No correctness, security, or best-practice concerns. It's a test-fixture value update only, with no impact on public SDK interfaces or consumers.

One minor observation (non-blocking): the BETA account config at line 85 still uses the old openai-reasoning-gpt-oss-20b-mtrl-mpg MPG name while its base_model is "mock-oss-test". If BETA is affected by the same drift, it may need a similar update — but that's out of scope for this PR and only worth checking if the BETA suite is failing too.

LGTM. ✅

[github-actions]

Claude finished @rsareddy0329's task in 22s —— View job

---

PR Review: Fix gpu integ test failure due to outdated MPG

• Read PR diff
• Review the change for correctness & conventions

Summary

This PR updates the PROD account (729646638167) model_package_group ARN from openai-reasoning-gpt-oss-20b-mtrl-mpg to mock-oss-test-mtrl-mpg in the MTRL trainer integ test config.

The change is correct and consistent with the surrounding PROD config, which already uses the mock-oss-test naming convention for base_model ("mock-oss-test") and existing_job_name ("mock-oss-test-mtrl-..."). The old MPG name referenced a model (openai-reasoning-gpt-oss-20b) that no longer matches what's used in that account, so this aligns the reference and fixes the outdated MPG that was causing the test failure. The PASSED test result in the PR description confirms the fix.

• ✅ Correctness: test-fixture value only; no impact on public SDK interfaces or consumers.
• ✅ No security, performance, or best-practice concerns.
• ✅ Scoped, minimal (+1/-1) change.

Non-blocking observation

The BETA account config (742774200982, line 85) still uses openai-reasoning-gpt-oss-20b-mtrl-mpg while its base_model is "mock-oss-test" — the same drift this PR fixes for PROD. If the BETA suite hits the same MPG failure, it may need a similar update. Out of scope for this PR, only worth checking if BETA is also failing.

LGTM. ✅