BeeTrove – OpenAI GPTs Open-Source Dataset

[robblbobbl]

https://beetrove.com/

[robblbobbl]

https://github.com/beetrove/openai-gpts-data - Thank you to the author for the effort and sharing!

[simonw]

Here's their CSV file with a scrape of the GPT catalog homepage from yesterday - 117 rows: https://github.com/beetrove/openai-gpts-data/blob/main/OpenA...

Here's a link to open that in Datasette Lite and apply a couple of facet filters to it (for author and list):

https://lite.datasette.io/?csv=https://github.com/beetrove/o...

This uses a custom SQL query to include their logos in the table as well: https://lite.datasette.io/?install=datasette-json-html&csv=h...

[andremafei → simonw]

This table with the logo is pretty cool, thanks for creating and sharing.

[sunbum]

Misleading title, should mention that its a dataset of CUSTOM Usergenerated GPT's.

[CharlesW → sunbum]

Agreed that it’s hard to parse, but “GPTs” is what OpenAI calls custom versions of ChatGPT.

https://openai.com/blog/introducing-gpts

[andremafei → sunbum]

The terms 'GPTs' and 'Custom GPTs' are used interchangeably by OpenAI.

[mritchie712]

Here's a sample of the data in a google sheet:

https://docs.google.com/spreadsheets/d/1woeD_wHiTvjK9hT7yMzF...

[tomComb → mritchie712]

I’m sorry that it doesn’t have the ‘top 10%’ version in a live Google sheet.

[CharlesW]

SYAC: This is content marketing with a CTA of "subscribe to André Mafei's newsletter", which is fine. The hook is that Mr. Mafei will send you some analysis of the data, none of which appear on the site or in the repo.

[TrueDuality → CharlesW]

The data is directly linked to in the repository if you'd like it:

Data: https://drive.google.com/drive/folders/1hUGnQ_AWeL2wi5UhUTt0...

Repo Source: https://github.com/beetrove/openai-gpts-data/tree/main/OpenA...

[andriym]

here is a data map of the descriptions, pretty cool: https://atlas.nomic.ai/data/andriy/openai-custom-gpt-data-ma...

[andremafei → andriym]

Awesome! Thanks for creating and sharing.