3 ms·Code embedding model with Matryoshka learning and quantization-aware training1 points by fzliu 2y ago