feat: Add model_utils and model_constants intorducing PROVIDER_NAMES and MODEL_INFO variables for dynamic updation of model data

[edwinjosechittilappilly]

This pull request introduces several significant changes to improve the handling of language models within the langflow project. The main updates include the addition of methods for building and retrieving language models, the introduction of a new class for model constants, and utility functions for model information retrieval. Additionally, a unit test has been added to verify the initialization of model constants.

ModelConstants Class

Introduced ModelConstants class in model_constants.py to hold model-related constants and solve circular import issues. This includes methods for initializing provider names and model information.

1. PROVIDER_NAMES:

   • This is a list that holds the names of model providers. It is populated with the display_name of each model from the MODEL_INFO dictionary.
   • The purpose of this list is to provide a quick reference to all the model providers available in the system.

2. MODEL_INFO:

   • This is a dictionary where each key is a model identifier, and the value is another dictionary containing details about the model.
   • The inner dictionary can include various attributes of the model, such as its display_name, configuration options, and other metadata.
   • This structure allows for easy access and management of model-related information.

3. initialize Method:

   • This static method is responsible for populating the MODEL_INFO and PROVIDER_NAMES variables.
   • It uses a delayed import to call get_model_info, which helps avoid circular import issues.

get_model_info Function

• Purpose: The get_model_info function is designed to retrieve information about all available models. This information is structured in a way that can be easily used to populate the MODEL_INFO dictionary.
• Functionality: It likely queries a database, reads from a configuration file, or interacts with an API to gather details about each model. The exact implementation would depend on your specific application setup.

How They Work Together

• When ModelConstants.initialize() is called, it fetches model information using get_model_info.
• The MODEL_INFO dictionary is populated with this data, providing a comprehensive view of all models.
• The PROVIDER_NAMES list is then derived from MODEL_INFO, offering a simple list of provider names for quick access.

This setup allows for efficient management and retrieval of model-related data, facilitating operations like model selection, configuration, and usage within the application.

Utility Functions for Model Information:

• Added get_model_info function in model_utils.py to gather inputs for all model components, ensuring that the necessary model information is available for building language models.

Unit Testing:

• Added a unit test in test_model_constants.py to verify that the ModelConstants class initializes correctly and that the provider names match the expected list.

Miscellaneous:

• Imported importlib in model.py to support dynamic module imports.

[edwinjosechittilappilly]

This Functionality will be later used in Dynamic LLM component and Simple agent component with Dynamic Inputs.