A place for developers to talk about building on Farcaster.

/dev

Has anyone written a script to preload chat gpt threads with very large context info. Eg multiple pdfs + code + docs? 

Or do I have to write it now lol

Cope dealer 🙏 🔲

Building interchain. Web3 UX flippening

GPT-4 context is like 8k tokens, not enough for much. The cool kids these days dump everything into a vector database, and then use terms from the prompt to max out the context window with relevant snippets of code and text etc, but that approach is still pretty limited.

Here's a great lesson from Andrew Ng on this topic:

Here's a great lesson from Andrew Ng on this topic:
https://www.deeplearning.ai/short-courses/langchain-chat-with-your-data/