How to look up the tokens based on the word representation?

[PhyllisH]

Hi~
I'm trying to use the ELMO replacing the original embedding lookup word representation in my model. And I see that you have a model in tensorflow https://tfhub.dev/google/elmo/2. But based on the example, I can only generate word representation and there is no api for the representation back to words. Could u help me with that?
Thanks

[matt-peters]

I'm not sure what you mean by an API to generate the representation back to words? To generate text from the LM?

[PhyllisH]

Yes. Can I use the token matrix as embedding matrix？

…

------------------ Original ------------------ From: Matthew Peters <notifications@github.com> Date: Sat,Jul 14,2018 2:40 AM To: allenai/bilm-tf <bilm-tf@noreply.github.com> Cc: PhyllisH <phyllis1sjtu@outlook.com>, Author <author@noreply.github.com> Subject: Re: [allenai/bilm-tf] How to look up the tokens based on the wordrepresentation? (#62) I'm not sure what you mean by an API to generate the representation back to words? To generate text from the LM? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub<#62 (comment)>, or mute the thread<https://github.com/notifications/unsubscribe-auth/AP8hD0c8ISCEUNbqD2q6yJ0SSzlCmGZFks5uGOmhgaJpZM4VIq56>.

[matt-peters]

We have a separate softmax matrix for prediction. You can get it as part of the tensorflow checkpoint, see these comments I just added to the README: https://github.com/allenai/bilm-tf/blob/master/README.md#are-the-softmax-weights-available