@lefterav Profile picture

Eleftherios Avramidis

@lefterav

Senior Research on Machine Translation/sign language, Natural Language Processing, Computational Linguistics, Fan of open source/free software and radio

Joined September 2011
Similar User
Tom Kocmi photo

@KocmiTom

Carolina Scarton photo

@carolscarton

Antonio Toral photo

@_atoral

Markus Freitag photo

@markuseful

Luisa Bentivogli photo

@luisabentivogli

Dr. Sheila Castilho photo

@_SheilaCastilho

Barry Haddow photo

@bazril

Dimitar Shterionov photo

@DShterionov

Fred Blain photo

@fblain

Jörg Tiedemann photo

@TiedemannJoerg

Carla Parra Escartín photo

@CParraEsc

MT Group at FBK photo

@fbk_mt

Rachel Bawden photo

@RABawden

fedegasp photo

@fedegasp1

Alina Karakanta photo

@karakanta

Help us break LLMs! The test suite sub-task will be included for the sixth time in the General MT Shared Task of the Conference on Machine Translation (WMT24). This year's theme is to reveal weaknesses of LLMs when translating #LLMs #EMNLP #NLProc| www2.statmt.org/wmt24/testsuit…


Eleftherios Avramidis Reposted

Happy to announce the WMT23 Metrics Shared Task! New this year amongst other changes: paragraph-level metric task for En-De & a new language pair: Hebrew-English. wmt-metrics-task.github.io #emnlp #NLProc #machinetranslation (1/n)

Tweet Image 1

Our #ACL2023 paper on "Neural #MachineTranslation for Translating Text to #SignLanguage Glosses" was named an outstanding paper! A result of the excellent MSc thesis of Dele @Yvan_Zhu at @TUBerlin with the invaluable help of Vera Czehmann. #NLProc @DFKI aclanthology.org/2023.acl-long.…

Tweet Image 1

traveling to the #ICASSP2023 #conference with to discuss at #SLTAT2023 the #design of our @DFKI prototypical #AI #SignLanguage #translation tool employing lightweight #AugmentedReality and smart headphones. dfki.de/en/web/researc…

Tweet Image 1

Eleftherios Avramidis Reposted

Meta evaluation of metrics usually does not get much attention. Dan's paper clearly shows that it is important to understand what your meta eval actually measures. Highly recommend to read! Really insightful work! Especially if you think about doing evaluation with LLMs.

LLM-based metrics like GEMBA predict many ties, but the way that ties should be handled in Kendall’s tau for meta-evaluating metrics has been a longstanding issue. We propose an update to the meta-evaluation methodology to handle ties. arxiv.org/pdf/2305.14324…

Tweet Image 1


The “Test suites” sub-task will be included for the sixth time in the General MT Shared Task of the Conference on Machine Translation (WMT23). Submission deadline: June 19th - more info: www2.statmt.org/wmt23/testsuit… #NLProc #EMNLP2023 #EMNLP #ACL


Really delighted that the #wearable Textile Game Controller of the @DFKI department of Interactive Textiles was presented by Esther Zahn and Hannah Fischer at the Interactivity Section of #CHI2023 in Hamburg

Tweet Image 1

Eleftherios Avramidis Reposted

Hello everyone! The deadline for submitting a paper to our workshop has been extended until April 24th. (Which means more time to work on great papers!) We're looking forward to all your contributions and to seeing you at @EAMT_2023 in Tampere! For info: sites.google.com/tilburgunivers…


Eleftherios Avramidis Reposted

Glad I am for the first time an instructor for two projects topics for 13 students of computer science at @TUBerlin_QU: "Towards automatic #SignLanguage translation" & "Fine grained evaluation of machine #translation". @DFKI @TUBerlin #NLProc

Tweet Image 1

My colleague was looking for citations on Bing Chat and got as a response a former non-published document of his with the same title but different made-up authors! #BingChat #ChatGPT

So I'm writing a grant proposal and using bing chat to find some related work.. suddently it cites one of my works, but with the wrong authors - with the same title and kinda niche topic (ai enabled biodiversity monitoring of surface waters). THE THING IS 1/2



Eleftherios Avramidis Reposted

Hello everyone! The Automatic Translation for Signed and Spoken Languages (AT4SSL) Workshop will be part of the @EAMT_2023, in Tampere, this year! If you are working on topics related to MT for Signed and Spoken Languages, have a look at our CfP: sites.google.com/tilburgunivers…


Excited that through our @DFKI #research project we can work on the automatic translation of sign languages. Our first experiments were shown with a technical report at the 7th Conference on #MachineTranslation, in #EMNLP2022. Still long way to go. #nlproc dfki.de/en/web/researc…

Tweet Image 1

Best long paper award at #EMNLP2022 on #tangrams!

Tweet Image 1
Tweet Image 2

Eleftherios Avramidis Reposted

Also check our last year’s metric task paper: aclanthology.org/2021.wmt-1.73/ . There are a lot of goodies: (7/n)


Eleftherios Avramidis Reposted

We welcome any additional comments and suggestions from the community regarding these evaluation plans and will take all such feedback into consideration. (6/n)


Eleftherios Avramidis Reposted

Please check out all details regarding both the main task and all subtasks: wmt-metrics-task.github.io (5/n)


Eleftherios Avramidis Reposted

For the Challenge Sets task we invite anyone to submit a new test suite and/or an analysis paper on metric behaviour for specific perturbations/phenomena similar to what we did last year! (4/n)


Eleftherios Avramidis Reposted

Similar to last year, we will not only evaluate the performance of metrics on the WMT submissions, but also augment the set of systems with additional MT systems generated by methods like sampling or MBR decoding. (3/n)


Eleftherios Avramidis Reposted

(a) Better Human Evaluation: Like last year, our primary human scores will be derived from expert-based MQM annotations. (b) We have 3 new subtasks: (i) QE as a Metric, (ii) Challenge Sets, (iii) MBR Decoding. (2/n)


Loading...

Something went wrong.


Something went wrong.