Inference
class is the primary facade for making requests to LLM providers in Polyglot.
It provides a unified interface for configuring providers, building requests, and executing inference operations.
Architecture Overview
TheInference
class combines functionality through traits:
- HandlesLLMProvider: Provider configuration and driver management
- HandlesRequestBuilder: Request construction and configuration
- HandlesInvocation: Request execution and PendingInference creation
- HandlesShortcuts: Convenient methods for common response formats