Productionizing diffusion models with Modal: QArt Codes deep dive скачать в хорошем качестве

Productionizing diffusion models with Modal: QArt Codes deep dive

In this video, Modal Developer Advocate Charles Frye walks through the details and architecture of a simple production application running on Modal: "QArt Codes", a QR code generator that uses diffusion models. Try it here: https://qart.charlesfrye.com Read the code on GitHub: https://github.com/charlesfrye/qart-c... Sign up for Modal for $30/month in free compute: https://modal.com/signup 00:00 - Intro and demo 04:35 - How do I connect a JS frontend to a Modal Python backend? 09:38 - How do I navigate a large Modal project? 10:45 - How do I run a development server on Modal? 13:26 - How do I deploy an existing FastAPI app on Modal? 17:45 - When should I use allow_concurrent_inputs on Modal? 19:01 - When should I use keep_warm or warm pools on Modal? 21:35 - How can I run a job queue on Modal? 26:58 - How can I monitor the outputs of ML models running on Modal? 31:28 - How do I use Modal Cls lookups? 34:38 - How do I find the code for a Modal Cls my app is using? 38:00 - How can I test code that runs on Modal? 41:28 - How do I use local and remote logs in Modal? 43:55 - How can I generate QR codes with diffusion models and ControlNets? 48:55 - How can I run end-to-end tests for Modal code? 52:09 - Outro