the other results, which use the OpenAI embeddings API got slightly worse. this also makes sense to me the API has likely changed a bit over the last two years they serve the models differently, use different quantization techniques etc. so there is a small distribution shift

→ View original post on X — @jxmnop