Dear @BraTS2024Participants , It has come to our attention that some users are experiencing issues being able to connect to the Synapse Docker registry, and unfortunately, the Synapse Platform team is not available at this time to address the issue. If you are one of these users, please follow this process instead: 1. push your image to another registry, such as [ghcr](https://docs.github.com/en/packages/working-with-a-github-packages-registry/working-with-the-container-registry#authenticating-with-a-personal-access-token-classic) 2. update the visibility permissions as needed. For example, all ghcr images are private by default 3. update your `mlcube.yaml` file so that `docker_image_name` contains the updated image name. 4. continue with the submission process and submit your MLCube config tarball to Synapse. For **Submission Name**, please indicate that your Docker image is available elsewhere such as "Docker on ghcr" so that I can be aware to manually check your submission. You will likely receive a follow-up email stating that your MLCube config submission is invalid due to a missing Docker image, but we will manually verify whether we can pull your image then update your final submission accordingly. Many apologies for the inconvenience 😔

Created by Verena Chung vchung
@cnb-service Don't mean to spam the group but since time is short I will. I have just uploaded to BraTS Africa my first submission but was hoping to double-check that everything is correct before I proceed with my other two. Got the "Your results will be posted to the leaderboard" message. And I assume that our submitted tar file no longer needs to have the same name as our docker image given we are uing ghcr?
@RamanK1103 , Yes, that will be the correct structure. To make it clear, let's say I have the following dummy model: ```text dummy_model ├── mlcube.yaml └── workspace ├── additional_files │ ├── model1.pt │ └── model2.pt └── parameters.yaml ``` After using MedPerf's convenience script, I then get the following tarball: ```text $ python scripts/package-mlcube.py --mlcube ~/Desktop/dummy_model --mlcube-types model --output ~/Desktop/model.tar.gz ...[truncated] ✅ Done! $ tar -tf ~/Desktop/model.tar.gz parameters.yaml mlcube.yaml additional_files.tar.gz ```
Hi @cnb-service , For my submission I am using several models during inference. All of these models are being stored (as saved tensorflow models) in the additional_files folder. They are then loaded directly in my run_inference function in my mlcube.py script. Concerning creating a tar file, can I just tar the ml cube folder as normal with the models nested inside the additional_files folder which will be nexted inside workspace folder? I am not uploading a model_weights file.
@ShadowTwin41 , Please submit your Docker images to the **BraTS 2024: Docker Images** queue rather than the MLCube queue.
Dear @cnb-service Thank you, I was able to submit again. However, I keep getting invalid submission. I am sure I'm submitting the docker first and the tar file next. I'm also sure that I give the same name to both submissions. What could I be doing wrong? Thank you!
@ShadowTwin41 , Apologies for that! You can now try again.
Dear @vchung, I'm having trouble submitting the Mlcube to the Meningioma Radiotherapy challenge. I have submitted first the docker and then the tar file. However, I'm getting our **MLCube submission (ID 9748163) is invalid. ** and I cannot submit once again because it says I reached my quota. Isn't it supposed for the invalid submissions to not count for the submission quota? Thanks in advance! Bet regards, André Ferreira
@juampablo That helped, thank you.
@aasandino regarding the "an image does not exist locally..." issue, remember to tag the image using a command like: ``` docker tag ghcr.io//name_segmentation:latest ``` where local_image_id can be found by typing "docker images" in terminal and name_segmentation is an arbitrary ID.
@aasandino , For your MLCube submissions, please add to the **Submission name** something like "model + additional data" or some other kind of label to indicate which model has/used additional datasets. Whatever name you choose for the Docker submission, please ensure you use the same one for the corresponding MLCube config submission :-)
@adap-sanyukta , Sorry to hear you're experiencing issues. So to confirm, when you do `docker images`, you can see `v1` listed under TAG for the image you are trying to push? e.g. ``` REPOSITORY TAG IMAGE ID CREATED SIZE ghcr.io/sanyuktaaap/adap-sanyukta v1 ... ... ... ```
Hello, For the BraTS-Path challenge we trained two models: 1) using challenge data 2) using Challenge data+additonal data. As per the challenge rules, if we used additional datasets, then we should submit two MLCubes: one using challenge data only, and one using challenge data plus the additional data. I was wondering if there are specific labels for the MLCubes for the submission. If so, how should I label these MLCubes? Thank you!
@vchung Running the command - **docker push ghcr.io/sanyuktaaap/adap-sanyukta:v1** gave me this error - **The push refers to repository [ghcr.io/sanyuktaaap/adap-sanyukta]. An image does not exist locally with the tag: ghcr.io/sanyuktaaap/adap-sanyukta** although the command **'docker images'** does show me the image in the list.
@adap-sanyukta , For NAMESPACE, please use your GitHub username. This will then push the image up to your GitHub profile at `https://github.com/YOUR_GITHUB_USERNAME?tab=packages`. From this page, you will be able to [update each package settings](https://docs.github.com/en/packages/learn-github-packages/configuring-a-packages-access-control-and-visibility#configuring-access-to-packages-for-your-personal-account) so that the image will be publicly accessible.
Hi Verena, For pushing docker images to ghcr, what do we enter in place of NAMESPACE in this command? - **docker push ghcr.io/NAMESPACE/IMAGE_NAME:latest** Earlier, for pushing it to synapse, we had in the form of - **docker push docker.synapse.org/SYNID/IMAGE_NAME:TA**G, but I have been getting an error saying 'denied: requested access to the resource is denied'. Could you please help me? Thanks, Sanyukta
Hi @RamanK1103 , At this time, there is no plan to extend the MLCube Submission phase due to the strict conference timelines, so the deadline is still set for tomorrow, August 7th, 23:59 UTC. If you would like, you can follow up with Ujjwal, one of the main BraTS organizers, at ubaid@iu.edu for additional details.
Thanks for the update Verena. Will the platform be deferring the deadline or is it sticking to 2359 UTC 7th Aug?

Issues with Pushing Docker Images page is loading…