Conversation
mufaddal-rohawala
left a comment
mufaddal-rohawala
left a comment
There was a problem hiding this comment.
🤖 AI Code Review
This PR fixes two example notebooks: corrects Session.boto_region_name to Session().boto_region_name (instance vs class access), updates MLflow code for v3.x API compatibility, and replaces direct boto3 invoke_endpoint calls with V3-native core_endpoint.invoke(). The changes are well-aligned with V3 SDK conventions.
| " order_by=['version_number DESC'],\n", | ||
| " max_results=1\n", | ||
| ")\n", | ||
| "\n", |
There was a problem hiding this comment.
Minor: The order_by parameter value 'version_number DESC' should be verified against the MLflow 3.x API. The documented field name for search_model_versions ordering is typically 'version_number', but some MLflow versions use 'creation_timestamp'. Please confirm this works with the MLflow version pinned in the notebook's dependencies to avoid a runtime error for users.
| "from sagemaker.serve.mode.function_pointers import Mode\n", | ||
| "\n", | ||
| "# Cloud deployment to SageMaker endpoint\n", | ||
| "# Note: 'dependencies' parameter is deprecated. You may see a deprecation warning.\n", |
There was a problem hiding this comment.
The comment says "Use configure_for_torchserve() for new projects" — is this an actual method available in the current SDK? If not, this could confuse users. Please verify the method name or remove the suggestion if it doesn't exist yet.
2 participants
Test PR