python - 当输入时我只有一个 ID 为 512 维向量的 json 文件时，如何处理三元组丢失？

Question

我有一个文件 json，其中包含从 InsightFace 和 MXNet 生成的向量，如下所示：

{"id": "huonght", "vector": "[-0.0591206625,-0.0562381446,...,0.0173285864,0.0403193906]}

{"id": "huonght", "vector": "[-0.0458422117,-0.0401753262,...,0.1048753187,0.0371856093]}

{"id": "huonght", "vector": "[-0.0422730967,-0.0223527066,...,0.0507882051,0.0381843373]}

...

我如何使用 Triplet Loss 将这些向量生成到集群中？