Dear @LISA2024_Organizers,
I had troubles creating a traball file based on `mlcube_docker_example.zip` ([Link](https://www.synapse.org/Synapse:syn55249552/wiki/627701)).
The `mlcube.yaml` file from the zip contains the following structure:
```
name: QC_testing
description: MLCube for running QC testing
authors:
- {name: "Your Name", email: "your.email@domain.com", org: "Your Organization"}
platform:
accelerator_count: 0
docker:
# Pre-built Docker image name.
image: my_image:latest
tasks:
test:
parameters:
inputs:
# Input directory containing images.
data: { type: directory, default: "data/" }
outputs:
# Output CSV file to be saved locally.
results: { type: file, default: "results/my_output.csv" }
entrypoint: python QC_testing.py --data {{inputs.data}} --results {{outputs.results}}
workspace:
# Local directory for inputs.
host: /App_mlcube
container: /workspace
```
When attempting to create the tarball, I followed the **"Preparing an MLCube for hosting" instructions as you advised ([Link](https://www.synapse.org/Synapse:syn55249552/wiki/627696))**.
However, the structure of the mlcube.yaml file described in those instructions differs significantly from the one you provided.
For example, the instructions require the task name to be "evaluate" instead of "test."
When I ran the `scripts/package-mlcube.py` with the necessary arguments, I received the following error message:
```
ERROR: Additional inputs must point to files in "additional_files"
```
Could you please provide guidance on how to correctly create the tarball file,
or offer a more detailed `mlcube.yaml` file that aligns with the hosting instructions?
Best regards,
Hyunwook Kim
Created by Hyunwook Kim wooks527 Hello,
Thank you for your question. As we mentioned here (https://www.synapse.org/Synapse:syn55249552/wiki/627701), mlcube_docker_example.zip is just an example:
**Important: The information provided is only an example. You should modify the contents of the Dockerfile and mlcube.yaml file according to the specific libraries and hardware configuration of your machine.**
So, you need to modify your MLCube configuration file to suit your specific requirements.
You can include a README file with instructions on how to run your model.
**Important: After packaging your models with Docker and MLCube, ensure they can run on a different system from the one it was built on to avoid any submission issues.**
Best regards,
LISA 2024 Challenge Organizers
Drop files to upload
Difficulties in Creating a tarball File page is loading…