Pivot-based transfer learning for neural machine translation between non-English languages
We present effective pre-training strategies for neural machine translation (NMT) using
parallel corpora involving a pivot language, ie, source-pivot and pivot-target, leading to a …
parallel corpora involving a pivot language, ie, source-pivot and pivot-target, leading to a …
Neural machine translation leveraging phrase-based models in a hybrid search
L Dahlmann, E Matusov, P Petrushkov… - arXiv preprint arXiv …, 2017 - arxiv.org
In this paper, we introduce a hybrid search for attention-based neural machine translation (NMT).
A target phrase learned with statistical MT models extends a hypothesis in the NMT …
A target phrase learned with statistical MT models extends a hypothesis in the NMT …
Learning from chunk-based feedback in neural machine translation
We empirically investigate learning from partial feedback in neural machine translation (NMT),
when partial feedback is collected by asking users to highlight a correct chunk of a …
when partial feedback is collected by asking users to highlight a correct chunk of a …
Document-level language models for machine translation
Despite the known limitations, most machine translation systems today still operate on the
sentence-level. One reason for this is, that most parallel training data is only sentence-level …
sentence-level. One reason for this is, that most parallel training data is only sentence-level …
Integrated Training for Sequence-to-Sequence Models Using Non-Autoregressive Transformer
…, J Rosendahl, W Wang, P Petrushkov… - arXiv preprint arXiv …, 2021 - arxiv.org
Complex natural language applications such as speech translation or pivot translation
traditionally rely on cascaded models. However, cascaded models are known to be prone to error …
traditionally rely on cascaded models. However, cascaded models are known to be prone to error …
LiLiuM: eBay's Large Language Models for e-commerce
C Herold, M Kozielski, L Ekimov, P Petrushkov… - arXiv preprint arXiv …, 2024 - arxiv.org
We introduce the LiLiuM series of large language models (LLMs): 1B, 7B, and 13B parameter
models developed 100% in-house to fit eBay's specific needs in the e-commerce domain. …
models developed 100% in-house to fit eBay's specific needs in the e-commerce domain. …
Word-based domain adaptation for neural machine translation
S Yan, L Dahlmann, P Petrushkov… - arXiv preprint arXiv …, 2019 - arxiv.org
In this paper, we empirically investigate applying word-level weights to adapt neural machine
translation to e-commerce domains, where small e-commerce datasets and large out-of-…
translation to e-commerce domains, where small e-commerce datasets and large out-of-…
Towards reinforcement learning for pivot-based neural machine translation with non-autoregressive transformer
…, J Rosendahl, W Wang, P Petrushkov… - arXiv preprint arXiv …, 2021 - arxiv.org
Pivot-based neural machine translation (NMT) is commonly used in low-resource setups,
especially for translation between non-English language pairs. It benefits from using high …
especially for translation between non-English language pairs. It benefits from using high …
Domain Adaptation of Foundation LLMs for e-Commerce
C Herold, M Kozielski, T Bazazo, P Petrushkov… - arXiv preprint arXiv …, 2025 - arxiv.org
We present the e-Llama models: 8 billion and 70 billion parameter large language models
that are adapted towards the e-commerce domain. These models are meant as foundation …
that are adapted towards the e-commerce domain. These models are meant as foundation …
[PDF][PDF] Language modeling and machine translation: improvements in training and modeling
G Yingbo - 2024 - www-i6.informatik.rwth-aachen.de
The field of statistical language modeling and machine translation has seen rapid developments
in recent years, with artificial neural networks taking center of the stage, dominating the …
in recent years, with artificial neural networks taking center of the stage, dominating the …