How process number in NLP task using embedding

In our task, we use embedding vectors to represent words and use these vector as the input of RNN. But there are many number in the texts and obviously there are many number have no embedding vector in the embedding model, how can we represent these number?

Are you sure that your embedding function doesn’t treat the numbers as words?

jpeg729 noreply@discuss.pytorch.org於 2018年1月12日 週五,下午9:53寫道:

jpeg729
http://post.spmailtechn.com/f/a/H06w-ywQG5Ti7yTltX-TCw~~/AAIPyQA~/RgRcOfE7P0QkaHR0cDovL2Rpc2N1c3MucHl0b3JjaC5vcmcvdS9qcGVnNzI5VwNzcGNYBAAAAABCCgAAu71YWgFANzU~
January 12

Are you sure that your embedding function doesn’t treat the numbers as
words?

Visit Topic
http://post.spmailtechn.com/f/a/A5Ntb0CVEAAk8yb56yLV_Q~~/AAIPyQA~/RgRcOfE7P0RTaHR0cDovL2Rpc2N1c3MucHl0b3JjaC5vcmcvdC9ob3ctcHJvY2Vzcy1udW1iZXItaW4tbmxwLXRhc2stdXNpbmctZW1iZWRkaW5nLzEyMjE2LzJXA3NwY1gEAAAAAEIKAAC7vVhaAUA3NQ~~
or reply to this email to respond.

To unsubscribe from these emails, click here
http://post.spmailtechn.com/f/a/ZI3AlP6-pxfAMPSR2HF-RA~~/AAIPyQA~/RgRcOfE7P0RtaHR0cDovL2Rpc2N1c3MucHl0b3JjaC5vcmcvZW1haWwvdW5zdWJzY3JpYmUvNDY1YTdmMjNkYWE1MTk5MWI2MWU1ODhlM2MyZWFmYTg1MDhmNjNmNGNkY2U0ODEyMTM3MzQ3ZDhjNzc1MzU4NlcDc3BjWAQAAAAAQgoAALu9WFoBQDc1
.

Yes,there are too many numbers in texts and many don’t contained in the
model ‘s vocal .