zelda-model-manager

To manage the Zelda and ASM model lifecycle, orchestrate dataset inventory, training runs, evaluations, registry updates, and deployment artifacts using a unified tool that integrates QA checks and model-mgr tooling.

You can convert and deploy GGUF or MLX artifacts using a local registry and model-mgr tooling, which handles the conversion process and serves the artifacts for Zelda and ASM model deployment.

The best way to coordinate ASM projects is through lifecycle orchestration that manages data prep, training runs, evaluations, and model serving, reducing coordination friction across teams.

Yes, a base model selection workflow is required. The lifecycle process mandates selecting a base model, running training, and performing QA checks before updating the registry or deploying artifacts.

Yes, the model-mgr CLI ties into registry configs and evaluation packs to ensure reproducibility and traceability throughout the ASM model training and evaluation lifecycle.

Zelda model deployment requires QA checks and registry updates to maintain reproducibility and traceability, ensuring that only evaluated and verified models are served as deployment artifacts.