Gradio 5 - Building a Quick Chabot UI for LangChain

[@jamesyoungerdds7901]

Sam - this is just awesome. I watch and absorb every video you make - just really grateful for your efforts and engagement in the community. It's such a moving target every week, but I make sure I tune in for everything you have to say. I really like your "code with me" videos as well. Please keep it up! 🙏

[@samwitteveenai]

Thanks for the comment, much appreciated!

[@epokaixyz]

Consider this your cheat sheet for applying the video's advice: 1. Use Gradio 5 to build a user interface for your LangChain chatbots. 2. Utilize Gradio's streaming functionality to display responses in real-time. 3. Experiment with system messages in LangChain to tailor your chatbot's personality and responses. 4. Leverage the ability to switch between different large language models (LLMs) for varied results. 5. Implement automatic memory to maintain conversational context throughout the chatbot interaction.

[@ИванИванов-б8у4и]

how to add image or file handling to the interface?

[@rorycawley]

Did the 'examples' parameter for ChatInterface just get broken? I now see the examples embedded into the chat panel rather than clickable buttons below the chat textbox demo_interface = gr.ChatInterface( examples=["hello"], fn=stream_response, textbox=gr.Textbox(placeholder="Send to the LLM...", container=False, autoscroll=True, scale=7), )

[@CaspersCuts]

Am I correct in thinking that memory of chat history is achieved by sending the full history with each query? If so, this uses more tokens right? And eventually concatenated history of messages would become greater than permitted query size / context window size? When that happens would the earliest messages drop off the memory or would we get error that query is now too large?

[@protovici1476]

I wonder if this and LightningAI are equal to get your programs to the public or even a paid service.

[@JurekOK]

what about "streaming graphs" that is, where the 2D plot of data needs to update at ~10fps ?

[@ajinkyapuranik9377]

Will this use up a bunch of tokens really fast?

[@samwitteveenai]

no it is just streaming the tokens. I asked for 500 words so we could see the streamig easily

[@ajinkyapuranik9377]

@@samwitteveenai I see! I guess I didnt catch that

[@PepitoGrillo-sq1mf]

I'm missing the second video of Agents