// MODEL OPTIMIZATION AND PROMPT SYNTAX TERM
Function Calling
When an AI model can identify that a user's request needs an external tool or function to be executed, and then generates the correct code to call that function.
TECHNICAL DEFINITION
Function calling enables large language models (LLMs) to generate structured JSON output that invokes external tools or APIs based on user prompts, facilitating interaction with databases, web services, or custom code.
BACKGROUND
In the field of artificial intelligence (AI), alignment aims to steer AI systems toward a person's or group's intended goals, preferences, or ethical principles. An AI system is considered aligned if it advances the intended objectives. A misaligned AI system pursues unintended objectives.
READ MORE ON WIKIPEDIASYNONYMS & ALIASES
- Tool use
- Tool calling
- API integration
- External function invocation
USAGE NOTE
Function calling extends LLM capabilities beyond text generation, allowing them to perform actions in the real world.
DEVELOPERS
Organizations developing technology related to Function Calling.
Pioneers in AI development, OpenAI provides models like GPT-3.5 and GPT-4 with advanced function calling capabilities, enabling them to interact with external tools and APIs.
Google offers robust function calling features through its Gemini models and Vertex AI platform, allowing AI agents to execute code, retrieve information, and interact with services.
Developers of the Claude family of models, Anthropic integrates 'tool use' capabilities, which serve as their equivalent of function calling, allowing the AI to invoke external functions for enhanced functionality.
Microsoft provides function calling through its Azure OpenAI Service, enabling enterprises to build AI applications that seamlessly integrate with their existing systems, databases, and APIs.
A leading framework for developing applications powered by LLMs, LangChain offers extensive support for agents and tools (function calling), enabling complex workflows and interactions with external systems.
Focusing on data frameworks for LLM applications, LlamaIndex often integrates with agent functionalities and tool use (function calling) to allow LLMs to interact with various data sources and external systems.
Hugging Face provides a platform and libraries for building, training, and deploying transformer models, including tools and resources that support the development of LLM agents capable of function calling.