3 ms·
I’m pretty sure they were using a high cost LLM to summarize, and for embeddings you only need Ada, which is orders pf magnitude cheaper.
by DanielVZ 4y ago
I’m pretty sure they were using a high cost LLM to summarize, and for embeddings you only need Ada, which is orders pf magnitude cheaper.