解决asked to gather along dimension 0, but all input tensors were scalars； will instead unsqueeze an

报错详情：

使用bert训练二分类模型

UserWarning: Was asked to gather along dimension 0, but all input tensors were scalars; will instead unsqueeze and return a vector.
....

runtimeErrror: expected long but get float

解决：

这是输入的类别个数不对，导致计算loss的时候这个错误。

下面的config.num_classes，如果是两类，那config.num_classes=2，这里我鬼使神差的写了1，然后各种查，也没发现问题。后来代码回退到上一个版本，完全可以正常运行，那肯定是本次添加的代码有问题，这里一行一行检查代码，发现了问题所在。

model = BertForSequenceClassification.from_pretrained(config.bert_name, num_labels=config.num_classes, output_attentions=False, output_hidden_states=True)