[BUG] Cannot Upload Scikit-Learn Models from the Tutorial

[stephenllh]

Is there an existing issue for this?

• I have searched the existing issues

Description

I want to upload a sklearn model, binary_classification_mock_credit_risk_sklearn.linear_model._logistic.LogisticRegression.sav from your tutorial.

I am seeing this

even though the stdout log seems to imply that the verification has passed:

Current Behavior

No response

Expected Behavior

I am expecting a success message, followed by an uploaded model in the "All Models" page

Steps To Reproduce

Build the Docker image and run the containers via docker-compose.yml as stated in the guide. Then, navigate to the AI model upload page.

Environment

- Operating System and Version: Ubuntu 22.04
- AI Verify Version: 0.9 (Docker tag version)
- Built with Dockerfile

Screenshots/ Code snippets

No response

Additional Context

No response

Possible Solution (Optional)

No response

[imda-benedictlee]

Hi @stephenllh, thank you raising this issue. Let us take a look and we will follow up with you once we have investigate the issue. In the meantime, will you be able to provide the logs from both the test-engine as well as from the apigw and portal?

[stephenllh]

Additional information: It worked if I run it on my local PC via WSL. But when I did it on a remote cloud server, this is the error.

[stephenllh]

Here are the logs extracted from the 2 containers:

test_engine_logs.txt
portal_logs.txt

[imda-benedictlee]

Hi @stephenllh, thanks for the logs. Unfortunately, I am unable to see any info for the test-engine_logs. Are you able to upload the test_engine_logs again?

For remote cloud server, there might be a configuration that you might need to modify. Let me double check on this and get back to you.

[stephenllh]

test_engine_logs.txt

Apologies. Here it is.

[imda-benedictlee]

Hi @stephenllh, thanks for the test-engine-logs. While we look into the issue in detail, you might like to try the solution below. As I understand the issue only occurs when using it in the cloud, we do have a potential solution that we have come up with for hosting AI Verify Docker in the Cloud. Below is a Docker Command that allows you to add the domain name / IP address of the cloud instance that you are running AI Verify Docker on in the command line argument for Docker Build.

bash docker-build.sh --build-arg PORTAL_URL=http://{domainname_or_ip} --build-arg WS_URL=ws://{domainname_or_ip} --no-cache

Where {domainname_or_ip} is the domain name or IP address of the host where you run the aiverify container instance, e.g. bash docker-build.sh --build-arg PORTAL_URL=http://25.255.11.123/ --build-arg WS_URL=ws://25.255.11.123/--no-cache

Do take note that though this is a potential solution, AI Verify is currently not officially supported for Cloud Instances. Do reach out to me here if you require additional assistance.

[stephenllh]

Thank you for the suggestion. Also, I just realized that cloud instance is not supported. Thank you for clarifying on that. I will try your solution to see if it works.

[imda-benedictlee]

Hi @stephenllh, may I know if the above solution suggested worked for you?

[hafeesyusuff]

Hi @imda-benedictlee I am having same issue as @stephenllh . In my case, I installed the App using the source code https://aiverify-foundation.github.io/aiverify/getting-started/source-code-setup/ (I installed Ubuntu on my windows 11). Is there anyway you can help me?

[imda-benedictlee]

Hi @hafeesyusuff, thanks for reaching out. Can I check at which part of the tutorial that you are stuck at? Also, may I know if you are using WSL or running AI Verify on Ubuntu in Virtual Machine, or via Docker?