在扩展识别模型的词典时是否可以继承之前的fc weight？ #14152
Unanswered
flow3rdown
asked this question in
Q&A
Replies: 1 comment 3 replies
-
|
有人能解答一下吗？ |
Beta Was this translation helpful? Give feedback.
3 replies
-
|
感觉不太行，不如试试单独训练FC层？ |
Beta Was this translation helpful? Give feedback.
-
语言模型扩展词向量一般都只是新建一些层，与之前的词向量合并。我不太清楚OCR是否可以这样做，直观上感觉保留之前的FC weight要比重新初始化会更容易训练一点，或者对数据的要求会更少一点？但我没具体实验尝试过所以想问下您是否有此方面的经验？ |
Beta Was this translation helpful? Give feedback.
-
|
这个不太一样吧，最后一层的FC类似于做一个分类操作，单独的直接合并FC层感觉不是一个有效的操作，这个我也没有实验过，不敢保证 |
Beta Was this translation helpful? Give feedback.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
扩展识别模型的词典大小后，FC weight的维度发生了变化，但是之前的词表是全部保留的，这种情况下是否可以保留之前的FC weight，新增扩展词的weight并与之前的FC weight进行合并？这样可行吗？
Beta Was this translation helpful? Give feedback.
All reactions