Models and datasets that were bulk-imported from the wider open-source ecosystem had their files stored correctly, but their detail pages weren't reading that content back out. Fixed the read path so READMEs, file listings, and Space file trees now show the real, already-stored content.