Using pretrained models for translations

[MickHardins]

Hello, I have a doubt regarding the use of the released pretrained models.

I have a marian server running the opus en-fr model (BPE). I'd like to test the model by translating some sentences of my choice.

According to model documentation, I have to send to the server preprocessed input. The file preprocess.sh usage is:

USAGE preprocess.sh langid bpecodes < input > output

While langid, input, and output are clear to me, I don't understand what should I pass as bpecodes. Can you please point me towards the right direction?

Thanks in advance

[jorgtied]

This should be the bpe-model file that comes with the translation model release.

[MickHardins]

This should be the bpe-model file that comes with the translation model release.

There are several files bundled with the model:

• decoder.yml
• source.bpe
• target.bpe
• opus..vocab.yml
• source.tcmodel

I tried both target.bpe and source.bpe. My understanding here is that I should use source.bpe since I'm encoding sentences in the source language. However when I send the text to the server using the websocket interfaces the server crashes.

[jorgtied]

Did you check the logfiles of your webserver that runs the service? Maybe some code is not available from the server? The source.bpe should be the model that you need.

[MickHardins]

The problem was a wrong decoder.yml file that was pointing to a different bpe file. Thanks for your help!