m3模型训练报错源码需要修改，不然会报"host_softmax" not implemented for 'Long'

[mfxss]

standard model训练正常，m3模型训练报错 FlagEmbedding/finetune/embedder/encoder_only/m3/modeling.py teacher_scores = teacher_scores.view(q_dense_vecs.size(0), -1).detach()后添加.float() 不然会报错 RuntimeError: "host_softmax" not implemented for 'Long'