Are you saving the match pairs somewhere? I imagine 1) there are a finite number of them, 2) doing an exact lookup in a DB first will be faster and easier than calling GPT3 and Pinecone every time, and 3) eventually GPT3 APIs will get pricey enough to make you think twice unless you're running your own instance on a cluster.
Comments
Are you saving the match pairs somewhere? I imagine 1) there are a finite number of them, 2) doing an exact lookup in a DB first will be faster and easier than calling GPT3 and Pinecone every time, and 3) eventually GPT3 APIs will get pricey enough to make you think twice unless you're running your own instance on a cluster.