A tool to visualize LLM applications before building them.
There are three main parts to the tool:
- Input
- Transformation
- Output
The idea is to think of LLM applications as a bunch of transformations applied to an input. The input is the raw data, the transformations are the LLM models, and the output is the result of applying the transformations to the input.
The tool allows you to see the input, the transformations, and the output in a single view. This is useful for understanding the flow of data through an LLM application, and building specifications irrespective of the framework of choice.
Motivation to build the tool is to track and understand the various states the building blocks of LLM applications enter into, thereby determining the state space of LLM applications.