Translated 0 sentences

[vemulapa...@gmail.com]

I have tried running a pretrained model of wmt16 on english-german. When i run it on the test set of europarl the sentences get translated. But when i give a subset say by doing head -n 500 eurparl-corpus. I get NFO: Translated 0 sents in 3.1948089599609375e-05 sec. Speed 0.0 sents/sec. Is there a format in which the input data has to be provided ?

[Rico Sennrich]

can you show the exact command you ran, and confirm that the file "europarl-corpus." exists?

if you're using the systems from http://data.statmt.org/rsennrich/wmt16_systems/ , a command like this should work (assuming some input_file):

cat input_file | ./translate.sh

On 10/04/2019 14:53, vemulapa...@gmail.com wrote:

I have tried running a pretrained model of wmt16 on english-german. When i run it on the test set of europarl the sentences get translated. But when i give a subset say by doing head -n 500 eurparl-corpus. I get NFO: Translated 0 sents in 3.1948089599609375e-05 sec. Speed 0.0 sents/sec. Is there a format in which the input data has to be provided ?

--
You received this message because you are subscribed to the Google Groups "Nematus Support" group.
To unsubscribe from this group and stop receiving emails from it, send an email to nematus-suppo...@googlegroups.com.
To post to this group, send email to nematus...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/nematus-support/60bcc0a0-2bad-4c74-a44e-d534f57b0356%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[vemulapa...@gmail.com]

When i run cat src-test.txt | ./translate.sh ,

INFO: Loading model parameters from file /home/harsha.v/Honors/ptModels/enDe/en-de/converted-model
INFO:tensorflow:Restoring parameters from /home/harsha.v/Honors/ptModels/enDe/en-de/converted-model
INFO: Restoring parameters from /home/harsha.v/Honors/ptModels/enDe/en-de/converted-model
INFO: Done
INFO: NOTE: Length of translations is capped to 200
INFO: Translated 80 sents
INFO: Translated 160 sents
INFO: Translated 240 sents
INFO: Translated 320 sents
INFO: Translated 400 sents
INFO: Translated 480 sents
INFO: Translated 560 sents
INFO: Translated 640 sents

But when i do head -n 200 src-test.txt | ./translate.sh,

INFO: Loading model parameters from file /home/harsha.v/Honors/ptModels/enDe/en-de/converted-model
INFO:tensorflow:Restoring parameters from /home/harsha.v/Honors/ptModels/enDe/en-de/converted-model
INFO: Restoring parameters from /home/harsha.v/Honors/ptModels/enDe/en-de/converted-model
INFO: Done
INFO: NOTE: Length of translations is capped to 200
INFO: Translated 0 sents in 3.5762786865234375e-05 sec. Speed 0.0 sents/sec

[Rico Sennrich]

whether you use 'cat' or head' makes no difference to the bash script ('./translate.sh'), nor Nematus itself.

you should double-check that there was no typo in the file name when you used 'head', and that the file was not accidentally overwritten or deleted before you executed the command.

On 10/04/2019 15:34, vemulapa...@gmail.com wrote:

--

You received this message because you are subscribed to the Google Groups "Nematus Support" group.
To unsubscribe from this group and stop receiving emails from it, send an email to nematus-suppo...@googlegroups.com.
To post to this group, send email to nematus...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/nematus-support/993a7aa1-73f7-44cd-9e66-31b48de6f0f0%40googlegroups.com.

[vemulapa...@gmail.com]

The file does exist. What i tried to show is it is giving translated 0 sentences when i give a smaller number of sentences which head -n 5000 did.

My main query is if there is a certain format that it expects with respect to line breaks etc. or the number of input sentences ?

To unsubscribe from this group and stop receiving emails from it, send an email to nematus...@googlegroups.com.