LLM Prompter

Options

Add system message

Specify whether a customizable system message is included in each prompt. This option is only available for chat models. Note that some chat models (e.g. OpenAI's o1-mini) do not support system messages.

Available options:

• None: No system message will precede the prompts.
• Global: A specifiable system message will precede each prompt.
• Column: Each prompt includes an individual system message specified in a column.

System message

The first message given to the model describing how it should behave.

Example: You are a helpful assistant that has to answer questions truthfully, and if you do not know an answer to a question, you should state that.

System message column

The column containing the system message for each prompt.

Prompt column

Column containing prompts for the LLM.

If there are missing values

Define whether missing or empty values in the prompt column and system message column (only applicable if the system message is provided via a column) should result in missing values in the output table or whether the node execution should fail on such values.

Available options:

• Output missing values: Rows with missing values will not be processed but are included in the output.
• Fail: This node will fail during the execution.

Output format

Choose between different output formats.

Available options:

• Text: Text output message generated by the model.

• Structured: When Structured is selected, the model is constrained to output data matching the defined structure. Each defined field will be extracted and added as a separate column in the output table. Not all models support structured outputs.

• JSON: When JSON is selected, the model is constrained to only generate strings that parse into a valid JSON object. Make sure you include the string "JSON" in your prompt or system message to instruct the model to output valid JSON when this mode is selected. For example: "Tell me a joke. Please only reply in valid JSON." Please refer to the OpenAI guide to see which models currently support JSON outputs.

Response column name

Name for the column holding the LLM's responses.

Output Structure

Settings for structured output extraction.

Target object name

The name of the target object. This helps the model understand what it is extracting. Invalid characters will be automatically converted to underscores.

Example: 'Person Info', 'Product Details', 'Sentiment Analysis'

Target object description

A description of what the target object represents. This provides context to the model about the overall extraction task.

Example: 'Information about a person including their name and age', 'Key details about a product'

Target objects per input row

Determines how many target objects are extracted for each input row.

Available options:

• Single: Extract exactly one target object with the defined columns per input row.
• Multiple: Allow the model to extract one or more target objects with the defined structure per input row. The input columns will be duplicated for each extracted object.

Input row ID column name

Name of the column that will store the original input row ID when multiple target objects are extracted per input row.

Output columns

Define the columns to extract from each prompt. Each column will be added to the output table. The model will be instructed to extract these columns from the input text.

    **Output columns:**

    - Column name : The name of the output column in the table.
    - Data type : The base data type of this column.

Available options:

• String: Text column.
• Number (Integer): Integer column.
• Number (Long Integer): Long integer column.
• Number (Float): Floating point number column.
• Boolean: True/False column.

  - Quantity : Determines if this column should contain a single value
  or multiple values (list) for each extracted item.
  
  **Available options:**
  
  - Single: Extract exactly one value of the defined type.
  - Multiple: Extract multiple values (as a list) of the defined type.
  
  - Description : A description of this column to help the model understand what to extract.
  

Views

This node has no views

Workflows

• No workflows found

Links

• No links available

Developers