Input Node

The input node allows you to pass data of various data types into a pipeline. While not always, inputs are usually the start of a pipeline.

Node Inputs

The input node does not have any node inputs.

Node Parameters

Type: The data type that the input will allow. Use the Type dropdown to signify the data type:

1. Text (default): raw text (e.g., a user message, text from an article, etc) that will appear as an editable text box.

2. File: The most common file types are: PDF, Doc, Docx, txt, xls, xlsx, csv, rtf, odt, tsv, ods, JPEG, PNG, etc. The file input type will appear as a field where users can upload a file.

3. Audio: Audio recording that you can record through the VectorShift platform.

Node Outputs

Based on the selected type, there are different node outputs for the input node

Input Type Text

1. Text: the inputted text.

   1. Type: Text

   2. Example usage: {{input_0.text}}

Input Type File

1. Processed_text: The inputted file processed into text

   1. Type: Text

   2. Example usage: {{input_0.processed_text}}

2. File: The inputted file

   1. Type: File

   2. Example usage: {{input_0.file}}

Input Type Audio

1. Audio: The inputted Audio

   1. Type: Audio

   2. Example usage: {{input_0.audio}}

Considerations

• The most common nodes the input node is connected to are the LLM node (as information in the prompt) and the knowledge base node (to query the knowledge base for information)

• You can expose input to end users by exporting the pipeline.

• Chat: The chat interface takes in one input of type Text and one output of type Text.

• Form: The form interface takes in an arbitrary number of inputs of any Type and an arbitrary number of outputs of any Type.

• Voicebot: The voice interface takes in one input of type Audio and one output of type Audio.

Example

The below example is a pipeline for chatting with a knowledge base. The input node, representing the user message, is connected to both the knowledge base (to query the knowledge base semantically) and to the LLM (so the LLM can respond to the user question).