摘要: 运行torch函数 torch.nn.functional.cross_entropy(pre, label, ignore_index=0)时报错,pre的shape为[ batch_size , n] label 的shape为[ batch_size]. 其中batch_size是batch的 阅读全文
posted @ 2020-01-08 22:15 东方神七 阅读(1581) 评论(0) 推荐(0)