samosx

Having a hard time with estimating how much GPU memory that LLM needs to serve it? What kind of GPUs to use and how many?

Wrote a blog post to demystify the process of GPU memory usage estimating.

Thanks for the feedback! All resources are namespaced right now, except there is an issue the notebook plugin where namespaces are indeed broken: https://github.com/substratusai/substratus/issues/193 Will get that fixed very soon.

Very simple prototype that uses stackoverflow API to get questions and chatGPT to answer them and store the results in postgres or sqlite using Peewee ORM.

OP here. You're right the solution is based on the existing documented full text search solutions. You can technically only sent the data you need indexed and the ID instead of sending all data. There is no native mechanism and firebase functions makes it as seamless as possible.

Feedback welcome, what feature would you want so this tool is helpful for you?