DOCSEARCHbyAlgolia
Submit Docs

LM Studio AI Model Catalog

repository·main·Indexed 21 days ago

https://github.com/lmstudio-ai/model-catalog

A collection of standardized JSON descriptors for Large Language Model (LLM) model files. It provides a unified way to describe model metadata, including size, architecture, file format, quantization format, authors, and providers.

Tokens
302
Snippets
0
Records
2
Agent score
24%

What's inside lmstudio-ai model-catalog

  1. Understand the model-catalog JSON descriptor format

    main

    The model-catalog uses standardized JSON descriptors to describe Large Language Model (LLM) models. Each model is represented by a <model_name>.json file located in the models/ directory.

    These descriptors capture metadata including:

    • Model size: (e.g., 7B, 13B, 30B)
    • Model architecture: (e.g., Llama, MPT, Pythia)
    • Model file format: (e.g., ggml)
    • Quantization format: (e.g., q4_0, q4_K_M)
    • Authors and resources: Links to academic papers or other metadata.
    • Files and providers: Information about available model files and their respective providers.

  2. Add a new model to the catalog

    main

    To contribute a new model to the catalog, follow these steps:

    1. Fork the repository and create a new development branch.
    2. Create a new JSON file for your model and place it in the models/ directory.
    3. Validate your JSON file against the schema using the provided tools: validate.py or createCatalog.py.
    4. Open a Pull Request (PR).
    5. Wait for the GitHub actions to complete successfully.

    Important: Do not modify catalog.json manually. The catalog.json file is automatically generated by a GitHub action that merges all individual .json files from the models/ directory.