mdagostino

I built a POC of the openai-realtime-console (https://github.com/openai/openai-realtime-console) in Streamlit for python folks like me who are terrible at javascript.

It's useful for poking around the events and the event payloads.

Next I'll integrate audio playback and then audio capture. Decoding the audio from the events is easy. Making streaming audio playback work with streamlit--not so much.

LLaMA is non-commercial

I wouldn’t say most—maybe a factor of 2. Getting the embedding is still an API call to an LLM.

This doesn’t make any sense to me: how does fine-tuning the embeddings save money? It seemed like the problem was having to make too many API calls to generate the embeddings in the first place.

Civis Analytics- Chicago, IL We spun out of the analytics shop from the Obama 2012 re-election campaign to tackle really hard data science problems for campaigns, non-profits, and companies. We're looking for both junior and senior software engineers and data scientists: http://www.civisanalytics.com/apply