Skip to main content
The LEAP Model Bundling Service is deprecated and no longer maintained. For new deployments, download GGUF files from Hugging Face and run them with llama.cpp.
The download command supports two modes of operation:
Download GGUF files for a model from the LEAP Model Library. Does not require authentication.
Arguments
  • model-name (required): Name of the model to download from the LEAP Model Library
Options
  • --quantization <quantization>: Quantization level to download for the model. You can find available quantization levels for each model in the LEAP Model Library. Defaults to Q5_K_M if not specified.
  • --output-path <path>: Directory to save the downloaded files. If not specified, auto-generates a directory name based on the manifest URL.
  • --overwrite: By default, if the JSON file already exists in the output path, the command will throw an error. With this flag, the existing file will be overwritten.
Behavior
  • Create an output directory based on the URL or according to --output-path if specified.
  • Download the JSON manifest file.
  • Download all model files referenced in load_time_parameters in the manifest
  • Update the local manifest to use relative paths to the downloaded files.
Examples
Manifest FormatThe JSON manifest should contain a load_time_parameters section with URLs to model files:
After download, the manifest is updated to use relative paths:
Error Cases
  • Invalid URL: Shows URL validation error
  • Manifest not found: Shows HTTP 404 error
  • Invalid JSON: Shows JSON parsing error
  • Model file download failure: Shows download error
  • Output directory creation failure: Shows file system error
  • Network issues: Shows connection error