Automating big tasks with AI

curl --location 'https://generativelanguage.googleapis.com/v1beta3/models/text-bison-001:generateText?key=' \
--header 'Content-Type: application/json' \
--data '{"prompt":{"text":"Give response in one word\ninput: Capital of India\noutput: New Delhi\ninput: The country with most Cricket world cups\noutput: Australia\ninput: 13/12/23 in a friendly format\noutput: December 13,2023\ninput: 25/1/2135 in a friendly format\noutput:"},"temperature":0.7,"top_k":40,"top_p":0.95,"candidate_count":1,"max_output_tokens":1024,"stop_sequences":[],"safety_settings":[{"category":"HARM_CATEGORY_DEROGATORY","threshold":"BLOCK_LOW_AND_ABOVE"},{"category":"HARM_CATEGORY_TOXICITY","threshold":"BLOCK_LOW_AND_ABOVE"},{"category":"HARM_CATEGORY_VIOLENCE","threshold":"BLOCK_MEDIUM_AND_ABOVE"},{"category":"HARM_CATEGORY_SEXUAL","threshold":"BLOCK_MEDIUM_AND_ABOVE"},{"category":"HARM_CATEGORY_MEDICAL","threshold":"BLOCK_MEDIUM_AND_ABOVE"},{"category":"HARM_CATEGORY_DANGEROUS","threshold":"BLOCK_MEDIUM_AND_ABOVE"}]}'

This curl can be translated using

Well, this uses the legacy model, PaLM-2 as the Gemini Pro API doesn’t work.

What’s helpful is if you provide an example of the Thunkable blocks that successfully make the API request shown in the curl command. That’s often the trickiest part of using an API.

That screenshot is impossible to read. :man_shrugging:

1 Like