This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
Next revision
Previous revision
non_native_speech_translation [2020/03/09 18:41]
obojar adding sebastian
non_native_speech_translation [2020/03/17 21:47]
obojar releasing test set
Line 53: Line 53:
 == Development Set == == Development Set ==
-  * A very minimal dev set: [[http://​ufallab.ms.mff.cuni.cz/​~bojar/​iwslt2020/​iwslt2020-nonnat-minidevset-v1.tar.gz|iwslt2020-nonnat-minidevset-v1.tar.gz]] (137 MB)+  * Devset v1: [[http://​ufallab.ms.mff.cuni.cz/​~bojar/​iwslt2020/​iwslt2020-nonnat-minidevset-v1.tar.gz|iwslt2020-nonnat-minidevset-v1.tar.gz]] (137 MB
 +  * Devset v2: [[http://​ufallab.ms.mff.cuni.cz/​~bojar/​iwslt2020/​iwslt2020-nonnat-minidevset-v2.tar.gz|iwslt2020-nonnat-minidevset-v2.tar.gz]] (149 MB, supersedes devset v1)
-Unfortunately,​ the dev set only illustrates **file formats**, including **expected output formats**.+The dev set illustrates **file formats**, including **expected output formats**.
-We will still try to extend ​the size of the dev set so that you can better assess your system quality during ​the test period.+Dev set v1 contained only a few sample files. Dev set v2 includes new files better illustrating ​the domain ​of the test set, but the reference translations are still not available for all files. 
 +== Test Set == 
 +  * Testset: [[http://​ufallab.ms.mff.cuni.cz/​~bojar/​iwslt2020/​iwslt2020-nonnat-testset.tar.gz|iwslt2020-nonnat-testset.tar.gz]] (270 MB) 
 +Please process all the files in the test set and produce formats as illustrated in the dev set.
 == File Format of ASR Candidates == == File Format of ASR Candidates ==
Line 130: Line 138:
 Discussion: <​iwslt-evaluation-campaign@googlegroups.com>​ Discussion: <​iwslt-evaluation-campaign@googlegroups.com>​
 +The non-native speech translation task is receiving support from the EU project [[http://​elitr.eu/​|ELITR]] (H2020-ICT-2018-2-825460).