add efficient_chat_with_ai

[merlinarer]

add an efficient chat_with_ai example discussed in here

[falkoschindler]

Thanks for the pull request, @merlinarer!

In order to keep it simple and more close to the original example, I updated the latter with the idea of incrementally building the UI from yours. Unfortunately, I'm always getting a rate limit when testing the UI. Can you or @rodja try if it works with your key?

[rodja]

I made some changes to display error messages and fixed getting the content from a chunk. But somehow the response is still not streamed.@merlinarer, could you look into this?

[merlinarer]

@rodja Have you tried my commit? I had already made it into streaming style.

[rodja]

@merlinarer Yes, in examples/efficient_chat_with_ai the streaming works. @falkoschindler tried to move the ideas to examples/chat_with_ai because we rather would like to have only one very well working example. Maybe you can have a look at the implementation?

[merlinarer]

@rodja @falkoschindler I abandoned ConversationChain, which would have some conflict with stream mode. Moreover, using config={"callbacks": [NiceGuiLogElementCallbackHandler(log)] is an alternative to set callbacks.

[falkoschindler]

Great! It works really nicely now. 🙂
Thanks a lot for the contribution, @merlinarer!