File Loader The file node allows you to load a static file into the workflow as a raw file or processed text from it (text extracted from the file). The node has two toggles at the top: Select / Upload File and File Name. If Select / Upload File is chosen, you have three options:
  1. Upload a file by clicking the upload button
  2. Use the dropdown to select a previously uploaded file to the VectorShift platform
  3. If switch the toggle from “Upload” to “Variable”, you can reference files from other nodes
If you switch the top toggle to File Name: you can enter the file name of a file within the VectorShift platform.

​
Node Inputs

If Select/Upload File selected:
  1. File: Select an existing file from the platform or upload a file for processing
    • Type: File
If File Name selected:
  1. File Name: The name of the file from the VectorShift platform from the files tab
    • Type: Text

​
Node Parameters

In the gear:
  1. Processing Model: The model that you want to use to extract the text from the file. The models available are: Default (Basic OCR), Llama Parse and Textract.
    • Use default if your files contain primarily text.
    • If your files contain complex tables, images, diagrams, etc., use Llama Parse or Textract (Note: additional costs will be charged).

​
Node Outputs

  1. File: The raw file, as-is
    • Type: File
    • Example usage: {{file_0.file}}
  2. Processed Text: The processed text extracted from the inputted file
    • Type: Text
    • Example usage: {{file_0.processed_text}}

​
Considerations

  • Click on the “+” button on the right of the node to create and connect the node to a semantic search node to enable semantic search on the file.

​
Example

The below example is a pipeline which answers questions about the Apple annual report (10-K).
  • Input Node: The query about the file.
  • File Node: Upload the file by clicking on the purple upload button (in this case the Apple annual report).
  • Semantic Search Node: Find semantically similar chunks to a question (input node) based on a file (file node).
    • Search Query:{{input_0.text}}
    • Documents for Search: {{file_0.processed_text}}
  • LLM Node: Answer a user query (input node) based on relevant information (semantic search node).
    • Prompt: {{input_0.text}} and {{semantic_search_0.chunks}}
  • Output Node: Display the LLM’s response
    • Output {{openai_0.response}}
File Example