Skip to main content
工作流应用

Large Language Model Search Optimization Node

A large language model node that supports surfacing search-related information.

Node Overview

Parameter NameDescriptionDefault Value
Model ConfigurationSelect an appropriate large language model and adjust its parameters.Qwen-Plus
Parameter ConfigurationParameters vary depending on the selected model. Includes: Temperature (controls output diversity) and Max Output Length (limits the maximum length of generated text).Default values retained
Search StrategyDifferent strategies affect search breadth and depth: Standard (standard search with single-round query rewriting), Enhanced (more rounds of rewriting and broader search sources), and Effect-Enhanced (further expands search sources).Enhanced
Display Search SourcesWhen enabled, search source information is combined with the user’s query and fed into the large language model.Enabled
Force Search ActivationWhen enabled, the large language model bypasses search intent detection and forcibly performs a search.Enabled
Badge DisplayWhen enabled, the large language model’s output includes badge styling (Style 1 or Style 2).Enabled, Style 1
Search ExpansionWhen enabled, enhances coverage for real-time information such as weather, stock prices, gold/silver prices, oil prices, and exchange rates.Enabled
System PromptProvides system-level instructions to the model—e.g., defining its role, task, or required output format.
User PromptThe interaction content between the user and the model, such as requests or instructions. Supports configuring prompt templates or inserting variables; the large language model processes and generates responses based on this configuration.${sys.query}
MemoryMemory functions equivalently to context—i.e., the model’s ability to retain prior conversation history in multi-turn scenarios. Supports two caching modes: Local Caching (retains only conversation history within this node; configurable number of turns) and Custom Caching (retains global context; requires selecting a context variable).Enabled, default settings retained
Output VariableSpecifies which parameter stores the node’s output result, enabling it to be referenced by subsequent nodes.Default value retained
Return ResultThis feature will soon be deprecated. It applies only when invoking the application via API and determines whether the node’s output content is returned.Default value retained

Configuration Example

Workflow structure: Start Node → Large Language Model Search Optimization Node → End Node