Integration into GPT-turbo and custom GPTs

Mikhail Hani

• November 07, 2023 20:54

In light of the new OpenAI updates and the enormous power of knowledge graphs with LLMs, I was wondering if a custom GPT using infranodus api could be used to take unstructured data stored in the GPT source database - convert them to KGs in infranodus and then provide them as vector replacements to be parsed through using the full power of GPT turbo.

1

• Official comment

  

  Dmitry Paranyushkin

  • February 12, 2024 10:31

  Hello Mikhail, how exactly would you want to feed back the knowledge graphs made with InfraNodus to GPT?
• 

  Mikhail Hani

  • February 12, 2024 20:25

  Hi Dmitry, 

  Thank you again for this incredibly powerful product.

  I was envisaging the creation of an infranodus customGPT which would take the prompt or entire conversation within chatGPT and create a knowledge graph with it. This graph would still need to be fine-tuned from the main website as usual, but this would allow regular chat GPT conversations to leverage the power of knowledge graphs and not limit the LLM response.

  The other application would be to link a fine-tuned knowledge graph from the website, and allow LLM output into the conversation - again removing the limitation of AI ouput. As the graph is tweaked, chatGPT's knowledge base would similarly be tweaked.

  My limited understanding of how to do this would be to utilise the "actions" api within the customGPT/agents section of openai's website. Something like this would be one of a kind for those without deep coding knowledge and would put infranodus on the map for many that would otherwise not know about it.

  Warm regards, Hani

  0

Please sign in to leave a comment.