Submission Failure (Music Demixing Validation)

I am able to run my code, demix, and receiver inference result. But the code continues to fail in the “Music Demixing Validation” step when I submit to AIcrowd. I am using a requirements.txt file to build the environment, and the build seems to work fine. The instructions mention an example file as the entry point, but it does not exist in any start kit. Rather, the starter kits’ readmes mention the MySeparationModel from as the entry point. Everything in my repo is configured to work with MySeparationModel, so I believe it should work in the submission. What am I missing here?