aws / aws/sagemaker-python-sdk

SageMaker Model Cards UI doesn't show Model Cards created using model package details

Offen
#5,080 0 Kommentare 0 Reaktionen 0 zugewiesene Personen Auf GitHub ansehen
component: Inference APIs and Interfaces type: bug
Vorherrschende Sprache
Python
Sterne
2.3k
Forks
1.3k
Ø Merge
1 T. 22 Std.
Gemergte PRs (30 T.)
35

Beschreibung

Describe the bug
When creating a model card using model package details, the model card is created and visible from the CLI but is not visible in the Model Cards section of the UI.

To reproduce
Create a model package details object like below. The code was obtained from the following [AWS article](https://aws.amazon.com/blogs/machine-learning/integrate-amazon-sagemaker-model-cards-with-the-model-registry/).

mp = ModelPackage.from_model_package_arn(
model_package_arn="arn:aws:sagemaker:us-east-2:000000000000:model-package/test-model-cards/7"
)

mc = ModelCard(
name="test-model-card-from-model-package",
model_package_details=mp
)

mc.create()
The output from the create function returns the ARN of the newly created model card i.e. arn:aws:sagemaker:us-east-2:000000000000:model-card/test-model-card-from-model-package' but does not show in the UI.

Expected behavior
The model card should be visible under Governance -> Model cards section of the SageMaker management console UI.

Screenshots or logs

Image

Model card can be found using CLI command aws sagemaker list-model-cards
list_model_cards_output

System information
A description of your system. Please provide:

SageMaker Python SDK version: 2.184.0
Framework name (eg. PyTorch) or algorithm (eg. KMeans): SKLearn
Framework version: 0.23-1
Python version: 3.10
CPU or GPU: CPU
Custom Docker image (Y/N): N
It should also be noted that we are experiencing this issue with custom algorithms deployed as ECR images.

Additional context
This is the same issue as the unresolved bug from 2023 which is already closed
https://github.com/aws/sagemaker-python-sdk/issues/4120

Beitragsleitfaden

Beitragsleitfaden öffnen

Rechercherichtung

Die gemeldeten Einstiegspunkte sind ModelPackage.from_model_package_arn, ModelCard(...), mc.create() und aws sagemaker list-model-cards; vergleiche die Felder der erstellten Model Card mit den Feldern, die die Model Cards UI auflistet. Prüfe Issue #4120 und verifiziere das Ergebnis in Governance → Model cards; erledigt ist die Aufgabe, wenn aus Model-Package-Details erstellte Karten dort erscheinen.

Vom Indexierungsmodell aus dem Issue-Text verfasst.

Bewertung

Tech-Stack
aws, machine-learning, python
Bereich
cloud, machine-learning
Issue-Typ
Bug
Schwierigkeit
4/5
Geschätzter Aufwand
3-5 Tage
Aktivitätsstatus
Veraltet
Klarheit
Größtenteils klar
Anfängerfreundlichkeit
35/100

Neue Issues direkt in Ihr Postfach

Eine kurze Übersicht über anfängerfreundliche GitHub-Issues.