我的Model.summary()的编号是多少?我无法清楚地理解.summary()在这里暗示什么

发布于 2025-01-23 02:37:10 字数 818 浏览 0 评论 0原文

我也知道嵌入层,双向LSTM和致密层。但是,我不清楚这里的编号实际上在做什么?那是我在同一层上的几次迭代吗?
因此,我的问题是:

  1. embedding_7 中的数字7是多少?
  2. 双向和 bidirectional_14 中的数字13和14是什么?

Layer (type)                 Output Shape              Param # 
==========================================================================
embedding_7 (Embedding)      (None, 300, 8)            19307592 
bidirectional_13 (Bidirecti  (None, 300, 256)         141312    
onal)  
bidirectional_14 (Bidirecti  (None, 256)              395264    
onal)
dense_7 (Dense)              (None, 9)                 2313      

=================================================================
Total params: 19,846,481
Trainable params: 19,846,481
Non-trainable params: 0

I know about the embedding layer, bidirectional LSTM and dense layers as well. However, I don't understand clearly that what are the numbering actually doing here? Is that for my several time iterations over the same layers??
So, my questions are:

  1. What is the number 7 in embedding_7?
  2. What is the number 13 and 14 in bidirectional_13 and bidirectional_14?

Layer (type)                 Output Shape              Param # 
==========================================================================
embedding_7 (Embedding)      (None, 300, 8)            19307592 
bidirectional_13 (Bidirecti  (None, 300, 256)         141312    
onal)  
bidirectional_14 (Bidirecti  (None, 256)              395264    
onal)
dense_7 (Dense)              (None, 9)                 2313      

=================================================================
Total params: 19,846,481
Trainable params: 19,846,481
Non-trainable params: 0

如果你对这篇内容有疑问,欢迎到本站社区发帖提问 参与讨论,获取更多帮助,或者扫码二维码加入 Web 技术交流群。

扫码二维码加入Web技术交流群

发布评论

需要 登录 才能够评论, 你可以免费 注册 一个本站的账号。

评论(1

耀眼的星火 2025-01-30 02:37:10

TensorFlow / keras层具有一个名称属性,必须是图形生成的唯一属性。如果您不提供“名称”关键字的(唯一!)名称,则TensorFlow / keras将为您生成一个名称(列表中的第一项,括号之前)。

为了使生成的名称与众不同,从0开始的增量数字附加了。因此,“密度_7”是网络中创建的第八个密集层。

Tensorflow / keras layers have a name property which has to be unique for graph generation. If you don't supply a (unique!) name with the 'name' keyword, tensorflow / keras will generate a name for you (the first term in the list, before the parentheses).

To make generated names unique, an incrementing number starting from 0 is appended. So 'dense_7' is the 8th Dense-layer the net has created in the graph.

~没有更多了~
我们使用 Cookies 和其他技术来定制您的体验包括您的登录状态等。通过阅读我们的 隐私政策 了解更多相关信息。 单击 接受 或继续使用网站,即表示您同意使用 Cookies 和您的相关数据。
原文