Skip to content

latent.simulators.conversation.models

Data models for conversation simulation.

Classes

SimulationConfig

SimulationConfig(max_turns: int = 30)

Configuration for conversation simulation.

Attributes: max_turns: Maximum number of turns before forced termination.

SimulationResult

SimulationResult(conversation: list[dict[str, str]], turn_count: int, success: bool, error: str | None = None, metadata: dict[str, Any] = dict())

Result of a single conversation simulation.

The conversation list contains dicts with role and content keys. These can be passed directly to Turn(**turn_dict) since all other Turn fields (timestamp, tool_calls, phase, metadata) are optional.

Attributes: conversation: List of turn dicts with role and content keys. turn_count: Number of exchange turns (user+assistant pairs). success: Whether the simulation completed without errors. error: Error message if the simulation failed. metadata: Additional metadata about the simulation.