Curated Data

VerbaGPT has a new exploratory feature, curated data! Now available to try for free on the web version ¹ of VerbaGPT.

Here is how:

1. Go to the link below. It will take you to a web version of VerbaGPT. Free, no sign up needed!
2. Choose “CURATED” from the [Data] tab and select a dataset from the dropdown (for example, “ML Data: iris”). We have some standardized datasets for you to try.
3. After selecting the dataset, VerbaGPT will show you some information about the data, so you can think of what question to ask. Or you can simply copy one of the sample questions provided in the description.
4. Go to the [Talk] tab, and ask the question!

Link: Access VerbaGPT Curated Data
Username: demo
Password: verbagpt

Important Note: For authentication to work properly, third party cookies must be enabled in your browser. This is not the case by default for Safari, Chrome Incognito Mode.

We designed this feature as an easy way to get people experimenting with large language models quickly and painlessly on some interesting datasets. Try it out and let us know what you think!²

1. VerbaGPT keeps data firewalled from large language models (LLMs), sharing only what is necessary to generate usable code. However, not all data is sensitive – and for some downloading software is a hassle. So, to get more people experimenting with using LLMs together with tabular data, we created an online version of VerbaGPT
2. The web version of VerbaGPT uses OpenAI’s GPT 3.5-turbo. It is faster but less accurate GPT 4. GPT 3.5 is still a great option to explore how LLMs can work with tabular data.