Model Catalog

What is a Model?

A model is a software program that uses algorithms/rules to make informed decisions, predictions, or generations based on a set of inputs without being given explicit instructions for every scenario. (e.g., ML models, Lookup tables, If-Else rules, LLMs, etc.)

A model typically includes one or more of the following components:

• Model file: Stores learned weights/parameters, lookup tables, tensors, and other important data to be used for initializing the model.
• Initialization Logic: Prepares the model or client for processing multiple inputs if needed.
• Scoring Logic: Code that applies the initialized model to provide inputs to generate/predict responses.

GGX supports the registration of various types of models:

• API-based models: Connect to externally hosted models like OpenAI, Gemini, etc., using APIs.
• Python-based models: Lightweight Python logic using various libraries or rule-based models.
• Custom models: Uploaded model files, including Scikit-learn models, NLP models like BERT, and any fine-tuned models.

Managing Models on the Platform:

The Model Catalog organizes all the registered models into customized groups at this centralized location, allowing easier tracking, monitoring, and model creation.

Registering a Model:

1. Click on Create button in Model Catalog.
2. Fill in important details like Name, Attributes (Output Type, Alias), Properties (Group, Permissible Purpose, Description, Approval Workflow).
3. Define Input Arguments along with their types and default values.
4. Select registered resources (like Model, Global Functions, Prompts, etc.) to use in model definition.
5. Select Input Type (API-Based, Python-based, or Custom registration).
6. Upload weights if required. Define model logic by writing code in Initialization Logic and Scoring Logic.
7. Select Training and Validation data(if any) used for building or testing the models.
8. Add notes, attach documentation if available in the Additional Information section.
9. Lastly, click on Save to complete the registration process.

The registered models can be evaluated in the Model Catalog or used in downstream objects (like RAG, Model, Pipeline, Reports, etc.).

Benefits of Model Registration:

• Automated tracking and recording of modifications with efficient version upgrades.
• Automatic detection of Permissible Purpose violations.
• Testing and Comparison with other registered models using custom and standardized validation kits.
• Enhances reusability across downstream applications and enables usage tracking with Lineage Tracking.
• Journey to production becomes more transparent and fully auditable, and production monitoring gets easier.
• Extract ready-to-productionize executable artifact.
• Fingerprinting of external API connectivity.
• Better collaboration for continuous model building and testing.