Explore generating example Qs for cold start problem

[pramitchoudhary]

Typically, on uploading a dataset we may not be sure what to ask. In order to come around this cold start problem can we use LLM + custom prompt to help guide the user with few sample Qs to get started.

[robinliubin]

import pandas as pd
df = pd.read_csv("<example.csv>")

from h2ogpt_client import Client
client = Client("https://gpt.h2o.ai", h2ogpt_key='***')
print(client.models.list())
text_completion = client.text_completion.create(model='h2oai/h2ogpt-4096-llama2-70b-chat')
print(text_completion.complete_sync(f"read below tabular data, create 10 sample questions from it, make it generic: ({df.columns})"))

[pramitchoudhary]

@robinliubin looking into this. Will update.

[pramitchoudhary]

Updated settings page,