Question about def init_embed_

I noticed that in the quantizer.py, you have a init_embed_ function which use data to init the embedding weight. In a distributed training environment, each rank has their own data, which leads to different init embeddings. Is this a problem?