dlgforge.pipeline.runner

Main generation and judge orchestration flows.

run(config_path)

Run synthetic conversation generation from a config file.

Parameters:

Name	Type	Description	Default
config_path	str	Path to a configuration file.	required

Returns:

Name	Type	Description
None	None	No value is returned.

Raises:

Type	Description
FileNotFoundError	Raised when validation or runtime requirements are not met.

Side Effects / I/O: - May perform network, model, or distributed runtime operations.

Preconditions / Invariants: - Callers should provide arguments matching annotated types and expected data contracts.

Examples:

>>> from dlgforge.pipeline.runner import run
>>> run(...)

run_judge_only(config_path)

Run judge evaluation on existing conversation artifacts.

Parameters:

Name	Type	Description	Default
config_path	str	Path to a configuration file.	required

Returns:

Name	Type	Description
None	None	No value is returned.

Raises:

Type	Description
FileNotFoundError	Raised when validation or runtime requirements are not met.
RuntimeError	Raised when validation or runtime requirements are not met.

Side Effects / I/O: - May perform network, model, or distributed runtime operations.

Preconditions / Invariants: - Callers should provide arguments matching annotated types and expected data contracts.

Examples:

>>> from dlgforge.pipeline.runner import run_judge_only
>>> run_judge_only(...)

run_multi_turn(cfg, output_paths, base_inputs, min_turns, max_turns, turn_count_distribution, turn_count_mean)

Run multi turn.

Parameters:

Name	Type	Description	Default
cfg	Dict[str, Any]	Configuration mapping that controls runtime behavior.	required
output_paths	OutputPaths	Filesystem path used by this operation.	required
base_inputs	Dict[str, Any]	Mapping payload for this operation.	required
min_turns	int	int value used by this operation.	required
max_turns	int	int value used by this operation.	required
turn_count_distribution	str	str value used by this operation.	required
turn_count_mean	float	float value used by this operation.	required

Returns:

Type	Description
Tuple[List[Dict[str, Any]], List[Dict[str, Any]], Dict[str, Any]]	Tuple[List[Dict[str, Any]], List[Dict[str, Any]], Dict[str, Any]]: Value produced by this API.

Raises:

Type	Description
Exception	Propagates unexpected runtime errors from downstream calls.

Side Effects / I/O: - May perform network, model, or distributed runtime operations.

Preconditions / Invariants: - Callers should provide arguments matching annotated types and expected data contracts.

Examples:

>>> from dlgforge.pipeline.runner import run_multi_turn
>>> run_multi_turn(...)

run_multi_turn_batched_async(cfg, output_paths, base_inputs, max_turns, batch_size, min_turns=1, turn_count_distribution='poisson', turn_count_mean=0.0, conversation_inputs_by_index=None) async

Run multi turn batched async.

Parameters:

Name	Type	Description	Default
cfg	Dict[str, Any]	Configuration mapping that controls runtime behavior.	required
output_paths	OutputPaths	Filesystem path used by this operation.	required
base_inputs	Dict[str, Any]	Mapping payload for this operation.	required
max_turns	int	int value used by this operation.	required
batch_size	int	Numeric control value for processing behavior.	required
min_turns	int	int value used by this operation.	1
turn_count_distribution	str	str value used by this operation.	'poisson'
turn_count_mean	float	float value used by this operation.	0.0
conversation_inputs_by_index	List[Dict[str, Any]] | None	List[Dict[str, Any]] | None value used by this operation.	None

Returns:

Type	Description
Tuple[Dict[str, Any], List[Dict[str, Any]]]	Tuple[Dict[str, Any], List[Dict[str, Any]]]: Value produced by this API.

Raises:

Type	Description
RuntimeError	Raised when validation or runtime requirements are not met.

Side Effects / I/O: - May perform network, model, or distributed runtime operations.

Preconditions / Invariants: - Callers should provide arguments matching annotated types and expected data contracts.

Examples:

>>> from dlgforge.pipeline.runner import run_multi_turn_batched_async
>>> run_multi_turn_batched_async(...)

persist_training_sample(output_paths, inputs, result, turns, raw_results)

Persist training sample.

Parameters:

Name	Type	Description	Default
output_paths	OutputPaths	Filesystem path used by this operation.	required
inputs	Dict[str, Any]	Mapping payload for this operation.	required
result	Any	Input value for this operation.	required
turns	List[Dict[str, Any]]	Conversation or message data used during processing.	required
raw_results	List[Dict[str, Any]]	Conversation or message data used during processing.	required

Returns:

Name	Type	Description
None	None	No value is returned.

Raises:

Type	Description
Exception	Propagates unexpected runtime errors from downstream calls.

Side Effects / I/O: - May perform network, model, or distributed runtime operations.

Preconditions / Invariants: - Callers should provide arguments matching annotated types and expected data contracts.

Examples:

>>> from dlgforge.pipeline.runner import persist_training_sample
>>> persist_training_sample(...)

persist_batched_training_samples(output_paths, base_inputs, conversations)

Persist batched training samples.

Parameters:

Name	Type	Description	Default
output_paths	OutputPaths	Filesystem path used by this operation.	required
base_inputs	Dict[str, Any]	Mapping payload for this operation.	required
conversations	List[Dict[str, Any]]	List[Dict[str, Any]] value used by this operation.	required

Returns:

Name	Type	Description
int	int	Value produced by this API.

Raises:

Type	Description
Exception	Propagates unexpected runtime errors from downstream calls.

Side Effects / I/O: - May perform network, model, or distributed runtime operations.

Preconditions / Invariants: - Callers should provide arguments matching annotated types and expected data contracts.

Examples:

>>> from dlgforge.pipeline.runner import persist_batched_training_samples
>>> persist_batched_training_samples(...)