LLM Tool

Large Language Models (LLMs) offer remarkable flexibility and can be seamlessly integrated within Dataworkz Agents. This integration supplies the Agent with an advanced suite of LLMs, enabling the execution of a diverse array of tasks tailored to specific use case requirements. The configurable nature of LLMs allows users to customize their functionality in multiple ways, enhancing their versatility. By leveraging the system prompt feature within the tool, users can fine-tune the behavior and responses of LLMs to meet exact specifications. This customization maximizes the potential of LLMs, ensuring they can adeptly handle various tasks, ranging from generating content to analyzing data, or delivering automated customer service. The adaptability and expansive capabilities of these models make them an indispensable component in many modern applications, providing substantial value by automating complex operations while maintaining high levels of accuracy and efficiency.

Pre-requisite

LLMs have to be configured on the Dataworkz platform using LLM configurator on the Home screen.

Set up

To setup an LLM tool that uses a Dataworkz RAG Application -

• Select AI Agents > Create a Tool > LLM

• Name: Provide a good name to the tool

• Description: Provide a useful description of the tool - usually this is the description of what this tool can be used for - e.g. this tool can be used to extract the information from any image passed on to it and give the results in a markdown format

• LLM Selection: Select the LLM to be used from the list

• System Prompt: Give a system prompt that the LLM needs to use along with a description of the inputs and the input variables. e.g. "The following url ${input_url} is an image. As a helpful assistant, you will extract the information from the image, analyze it and provide the output in the following format: raw_data - the complete extracted text, summary - summary of the extracted data. The out should be in markdown format." Here input_url becomes the Input Parameters in the next step

• Input Parameters: The input parameters defined in the previous step appear here and you should adjust the type and the description of the parameters.

• Output: The output of the tool is pre-determined