Building AI-powered apps with Replicate and Streamlit

In this video, Zeke, a Software Engineer from the Replicate team, joins us for a tutorial on using open-source AI models hosted on Replicate in a Streamlit app.

Don't feel like watching a video? Here are some other ways to explore Replicate and the demo app:

⏰ Timeline
0:00 About Zeke
0:49 Introduction and overview of Replicate
3:09 Available models on Replicate (image generation, text generation, etc.)
4:12 Streamlit demo using Replicate
7:29 How to get started with Streamlit
13:56 How to get started with Replicate
17:40 Tips for using Replicate (data persistence, webhooks)
20:23 Model performance and streaming capabilities
26:17 Deployments for production-scale applications
28:14 Tips for building scalable apps with Replicate
31:27 Searching for models on Replicate
32:42 Multimodality support in Replicate
34:51 Hardware options and computational power
36:44 Thoughts on the potential of language models
39:10 Demo app/code walkthrough
45:12 Conclusion

🎬 Credit

[andrebadini]

The Streamlit community would greatly benefit from a feature that allows for isolated cookie saving for different users. Has this been considered by the development team?

[caiquecober]

can i use public models on ollama in replicate?
That are finetuned to my usecase as it was my private version on replicate?

[rodrigobogado]

Recien llegue, lo vere antes de dormir. Gracias por el video!!!