知识卡片
训练数据方向与示例必须一致
本章在Multi30k与OPUS Books、英译德与德译英之间切换叙述。运行时需固定真实数据集、语言方向及字段,不能将英语源词表和德语目标训练后,直接把德语例子按另一套旧Field反解并当同一次实验。
这张卡片看不懂?
这张卡片已通过人工审查,不再接受失败反馈。
已有反馈 · 0 条
还没有人提交反馈。
在整书笔记中的位置
原书出处 · 1 处
- · 查看原书依据
9.2.3 Transformer translation Transformers are suited for many tasks. “Attention Is All You Need” showed off a transformer that achieved better translation accuracy than any preceding approach. Using torchtext, you will prepare the Multi30k dataset for training a transformer for German–English translation, using the torch.nn.Transformer module. In this section, you will customize the decoder half of the Transformer class to output the self-attention weights for
参考来源
- 位置:《Natural Language Processing in Action, Second Edition》第09章《可堆叠的Transformer》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Natural Language Processing in Action, Second Edition/_epub-src/OEBPS/Text/chapter-9.html)
- 结论依据:摘录中关于训练数据方向与示例必须一致的定义、比较或条件支撑卡片;原书表述不严谨的范围已在覆盖记录逐项限定,不把案例结果外推为普遍结论。
- 原始内容:9.2.3 Transformer translation Transformers are suited for many tasks. “Attention Is All You Need” showed off a transformer that achieved better translation accuracy than any preceding approach. Using torchtext, you will prepare the Multi30k dataset for training a transformer for German–English translation, using the torch.nn.Transformer module. In this section, you will customize the decoder half of the Transformer class to output the self-attention weights for