Using LLM to Expand Data

Tsubasa Kato
1 min readOct 14, 2023

--

Addition to the article I wrote 3 days ago, I’ve made a video of the Python script that enables data to be expanded using a LLM. In this example, I am using Stability AI’s Stable LM 3B to expand data from a file containing Wikipedia article. It then sort of inflates the data using the Stable LM 3B.

This example shows how small subset of data can be used to be a source to expand its data by feeding it to a LLM. This can be useful for offline data expansion, for example, in ideation of a product, making a story or article from several words etc. It can also be applied to brainstorming tools.

Update 11/4/2023:

I’ve uploaded a multi-thread version of data expander on GitHub. It is much faster compared to the single thread version I uploaded earlier.

https://github.com/stingraze/llm-seed-url-generator/blob/main/experimental/stabilityai-data-expander-multi-thread.py

Follow my Twitter / X at: https://www.twitter.com/_stingraze

My LinkedIn: https://www.linkedin.com/in/tsubasakato/

--

--

Tsubasa Kato
Tsubasa Kato

Written by Tsubasa Kato

Japanese Search Engineer. Likes Quantum Computers and handheld computers.

No responses yet