Implementation:Ucbepic Docetl Directive OperatorFusion

Knowledge Sources	Ucbepic_Docetl
Domains	Pipeline_Optimization, LLM_Operations
Last Updated	2026-02-08 00:00 GMT

Overview

Concrete tool for combining two sequential operations into a single operation to reduce LLM processing costs provided by the DocETL reasoning optimizer.

Description

The OperatorFusionDirective class combines two sequential operations into a single operation to reduce LLM processing costs by avoiding duplicate document reads and API calls. It can fuse Map+Filter, Map+Map, and Map+Reduce patterns into a single combined operation with a merged prompt that performs both tasks simultaneously.

Usage

The MOAR agent applies this directive when there are two sequential LLM operations processing the same document keys and the goal is to optimize cost by combining them into one operation. Two consecutive operations must be specified as target operators.

Code Reference

Source Location

• Repository: Ucbepic_Docetl
• File: docetl/reasoning_optimizer/directives/operator_fusion.py
• Lines: 1-329

Signature

class OperatorFusionDirective(Directive):
    name = "operator_fusion"
    description = "Combines two sequential operations into a single operation to reduce LLM processing costs."

    def check_applicability(self, ...) -> Tuple[bool, str]: ...
    def apply(self, ...) -> Tuple[List[Dict], List[Dict], str, dict]: ...

Import

from docetl.reasoning_optimizer.directives.operator_fusion import OperatorFusionDirective

I/O Contract

Inputs

Name	Type	Required	Description
op_config	Dict	Yes	Operation configuration to transform
pipeline_ops	List[Dict]	Yes	Full pipeline operations list
op_idx	int	Yes	Index of target operation
dataset_descriptions	Dict	Yes	Dataset schema descriptions

Outputs

Name	Type	Description
new_ops	List[Dict]	Transformed operation configs
new_steps	List[Dict]	Updated pipeline steps
explanation	str	Human-readable description of changes
metadata	dict	Additional metadata about the transformation

Usage Examples

# Directives are typically invoked by the MOAR agent automatically
# Example of manual invocation:
from docetl.reasoning_optimizer.directives.operator_fusion import OperatorFusionDirective

directive = OperatorFusionDirective()
applicable, reason = directive.check_applicability(op_config, pipeline_ops, op_idx, dataset_descriptions)
if applicable:
    new_ops, new_steps, explanation, metadata = directive.apply(op_config, pipeline_ops, op_idx, dataset_descriptions)

Related Pages

• Environment:Ucbepic_Docetl_Python_Runtime
• Environment:Ucbepic_Docetl_LLM_API_Keys