I'm working on a restaurant similarity search engine and leveraging semantic search to deliver more accurate results. I was curious to see how different embedding models would perform on this task. You can find the notebook with the code and the results here
I've also written a blog post about this project.