I’m Using chatgpt Completions API and install link reader plugin in my OpenAI Account.
So, my question is using this Completions API can chatGpt use link reader plugin to get link data that provided as a request?
The ChatGPT Plugins ecosystem cannot be used with the API, the two are separate entities, the functionality is similar but not compatible.
Thank You for response.
Know any other solution or integrate chatGpt plugin?
You can do anything a Plugin can do with the API and it’s even simpler now with Functions, as the model is now tuned to give JSON results back you can readily read back the replies from the AI and make calls based on that, gather data and feed that back into the prompt for further processing.
Problem is I have large document file and i have summarize data all at once and this completions API have limitations of prompt size
For text corpuses that are larger than the prompt size, you can use embeddings to encode the text into a vector database. You then run a query on that database to pull back information that is relevant and then run the LLM model on that as context, along with the original query. That is in essence what a chat with pdf style plugin does.
LangChain or other similar projects have some code to get you started on this sort of thing
https://python.langchain.com/en/latest/modules/chains/index_examples/summarize.html
1 Like
Looking for something similar, I want to summarize text from a website, and linkreader works so nice but I can’t do this using openai, the html page is huge.
Also breaking it down to chunks doesnt work because many of the data on a webpage is just noise.