CLI Command Group

moro release

Export models into standalone quantized runtimes (GGUF), sign cryptographic Software Bills of Materials (SBOM), and deploy directly into Ollama or Docker.

Subcommands

1. `moro release export`

Fuses the trained LoRA adapter weights with the base model, applies quantized GGUF packaging, and generates a signed provenance manifest.

moro release export --run-id run_01 --version v1.0.0 --quant Q4_K_M
  
Flag Type Default Description
--run-id String Required Training run identifier from runs/ directory.
--version String v0.1.0 Semantic version tag for this release.
--quant String Q4_K_M Quantization type: Q4_K_M, Q5_K_M, Q8_0, or F16.

2. `moro release deploy`

Registers the exported release bundle directly into local serving runtimes:

# Deploy to local Ollama runtime
moro release deploy --target ollama --version v1.0.0 --name support-model

# Deploy as an isolated Docker container
moro release deploy --target docker --version v1.0.0 --port 8080
  

3. `moro release verify`

Audits the cryptographic signatures and SHA-256 hashes inside the model's SBOM:

moro release verify --version v1.0.0