nat.experimental.test_time_compute.functions.ttc_tool_orchestration_function#

Attributes#

logger

Classes#

TTCToolOrchestrationFunctionConfig

Configuration for the TTCToolOrchestrationFunction, which is used to orchestrate multiple functions.

Functions#

register_ttc_tool_orchestration_function(config, builder)

Registers an TTC-based orchestration function that:

Module Contents#

logger#
class TTCToolOrchestrationFunctionConfig(/, **data: Any)#

Bases: nat.data_models.function.FunctionBaseConfig

Configuration for the TTCToolOrchestrationFunction, which is used to orchestrate multiple functions.

Create a new model by parsing and validating input data from keyword arguments.

Raises [ValidationError][pydantic_core.ValidationError] if the input data cannot be validated to form a valid model.

self is explicitly positional-only to allow self as a field name.

augmented_fns: list[nat.data_models.component_ref.FunctionRef] = None#
search_strategy: nat.data_models.component_ref.TTCStrategyRef | None = None#
editing_strategy: nat.data_models.component_ref.TTCStrategyRef | None = None#
scoring_strategy: nat.data_models.component_ref.TTCStrategyRef | None = None#
selection_strategy: nat.data_models.component_ref.TTCStrategyRef = None#
async register_ttc_tool_orchestration_function(
config: TTCToolOrchestrationFunctionConfig,
builder: nat.builder.builder.Builder,
)#
Registers an TTC-based orchestration function that:

  1. Instantiates all relevant strategies (search, editing, scoring, selection).

  2. Accepts a ToolUselist, converts each item to an TTCItem, optionally runs search/editing.

  3. Calls the correct augmented_fn per item using name=tool name.

  4. If configured, runs scoring and selection on the result.

  5. Returns a new ToolUselist with each output set.