We are currently using a custom setup based on a manually created manifest.json and supporting logic.
This approach was designed for simplicity, but as requirements grow more complex, it demands more manual configuration and maintenance.
Therefore, similar to how Hugging Face manages model repositories, we plan to transition to a Git-based repository structure for the model cache.
This would allow us to take advantage of versioning, remote synchronization, and distributed collaboration features inherent to Git (or XET or any kind of VCS).
Although the exact design and implementation plan have not yet been finalized, this is an important direction that we intend to pursue in the future.
We are currently using a custom setup based on a manually created manifest.json and supporting logic.
This approach was designed for simplicity, but as requirements grow more complex, it demands more manual configuration and maintenance.
Therefore, similar to how Hugging Face manages model repositories, we plan to transition to a Git-based repository structure for the model cache.
This would allow us to take advantage of versioning, remote synchronization, and distributed collaboration features inherent to Git (or XET or any kind of VCS).
Although the exact design and implementation plan have not yet been finalized, this is an important direction that we intend to pursue in the future.