Dario Stojanovski

Applied Scientist at Microsoft, PhD from LMU Munich

[Alexandra Chronopoulou @alexandraxron]

📣 If you are still at Dubrovnik #EACL2023 tomorrow, I am presenting our paper “Language-Family Adapters for Low-Resource Multilingual Neural Machine Translation” at #LoResMT oral session 2 at 11:55am!☀️

Joint work w/ @StDario1, A. Fraser

✏️ https://t.co/udkZuRCjN2

[Alexandra Chronopoulou @alexandraxron]

Can we build models that benefit from transferring knowledge between related domains?

In our #NAACL2022 paper w/@mattthemathman+ @JesseDodge, we propose an efficient method of adapting PLMs to multiple domains.

Paper: https://t.co/dZukKTgAHD
Blog: https://t.co/skJ07ZH7ZR
🧵1/6

[Alexandra Chronopoulou @alexandraxron]

Happy to share that our paper "Improving the Lexical Ability of Pretrained Language Models for Unsupervised Neural Machine Translation" got into #NAACL2021!🎉Joint work w/ @StDario1 and Alex Fraser.

Here is a summary thread👇

[Mathias Müller @bricksdont]

Out now: my thoughts and recommendations on machine translation evaluation. Would love to hear _your_ thoughts and recommendations! Any feedback is welcome. #neuralempty

https://t.co/CG0eDSf5No

[Dario Stojanovski @StDario1]

We show context NMT accuracy on pronoun translation challenge sets can be manipulated w/ adversarial attacks and propose template test sets to evaluate CR in MT. Results show CR is poorly handled in MT. Also proposed a data augmentation method for better pronoun translation(2/2)

[Dario Stojanovski @StDario1]

Can NMT handle coreference? Presenting our #coling2020 paper "ContraCAT: Contrastive Coreference Analytical Templates for MT" on Dec 10 at 17:00 CET! Joint work w/ @benno_krojer, Denis Peskov and Alex Fraser.

Data: https://t.co/9hbTu1GsB7
Paper: https://t.co/UOzZFvsJSC (1/2)

[Alexandra Chronopoulou @alexandraxron]

Are you in #emnlp2020 and interested in leveraging monolingual data for machine translation in low-resource scenarios? If yes, come take a look at our poster "The LMU Munich System for the WMT 2020 Unsupervised Machine Translation Shared Task" in Room A, at 12:00-13:30 CET 😄

[Alexandra Chronopoulou @alexandraxron]

We achieved 1st place in the #WMT2020 Unsupervised MT task! Core elements: (twisted) MASS pretraining, online BT on monolingual data 📚+ translation loss on pseudo-parallel data📕📗from unsup. SMT(+BPE drop)

Joint work w. @StDario1, V.Hangya, A.Fraser
📄https://t.co/zKVNmO3RDk

[Alexandra Chronopoulou @alexandraxron]

📜Pre-print of the paper: https://t.co/EhpzFonZFH

🤖Code: https://t.co/TLPFA9Ary0

see you all virtually at @emnlp2020 🥳

[Alexandra Chronopoulou @alexandraxron]

Excited to share that our paper "Reusing a Pretrained Language Model on Languages with Limited Corpora for Unsupervised NMT" is accepted in the main conference of @emnlp2020!🥳Thanks so much to my co-author @StDario1 and advisor Alex Fraser!Arxiv link + code coming up soon!(1/4)

[Aleksandar @mihajlovskia]

Прва моја Андроид апликација која би требало да го олесни плаќањето паркинг преку СМС. Опции кои ги нуди моментално апликацијата се:
- Oбележани зони за Скопје, Куманово, Струмица
- Внесување регистарска таблица
- Праќање СМС
- Потсетник за завршен паркинг
https://t.co/nFp9WsELhk

[SpaceX @SpaceX]

Falcon Heavy side cores have landed at SpaceX’s Landing Zones 1 and 2.