In this repository, I will put the prompts I am using to create a dataset in the style of Stanford's Alpaca in order to create a Generative UI tool.
The final goal is to create a generative model capable of making Tailwind UI's with human-readable prompts. The final model will be made on top of Mistral AI models and of course, it will be open source with commercial usage permissions.
The Alpaca Style is usually a pair of instruction and response and the final dataset will look like this:
| Instruction | Response |
|---|---|
| A sleek pricing page for a SaaS. | <html>...</html> |
The first approach was to create prompts, then codes. I personally think it is still in good shape, but only if you have access to OpenAI's API and automate the procedure. Otherwise it is not really a reliable way. For now, we need another approach to acquire our needed data.
In the new approach, we're looking for an Alpaca like dataset generated by ChatGPT itself. It may be a little tricky, but at the end of the day it might be the best choice we've ever had.
Since the new approach worked perfectly but not as good as expected, the final decision was to use poe.com bots to creat our desired dataset. A super-prompt will be written for poe.com and then it'll be used to generate our dataset.
- Generate prompt generator prompt.
- Prompts were great but not basically what we wanted. We need to rethinkg the procedure.
- Enhance prompt generator to create UI components only.
- Generate code generator prompt.
- Define a new approach.
- Generate new approach prompt.
- Generate a super-prompt for poe.com
- Generate component generator prompt.
- Make the basic dataset.
WILL BE COMPLETED