Zobrazeno 1 - 10
of 36
pro vyhledávání: '"Libovický, Jindrich"'
Autor:
Helcl, Jindřich, Kasner, Zdeněk, Dušek, Ondřej, Limisiewicz, Tomasz, Macháček, Dominik, Musil, Tomáš, Libovický, Jindřich
This paper presents teaching materials, particularly assignments and ideas for classroom activities, from a new course on large language models (LLMs) taught at Charles University. The assignments include experiments with LLM inference for weather re
Externí odkaz:
http://arxiv.org/abs/2407.19798
Autor:
Libovický, Jindřich, Helcl, Jindřich
We present three innovations in tokenization and subword segmentation. First, we propose to use unsupervised morphological analysis with Morfessor as pre-tokenization. Second, we present an algebraic method for obtaining subword embeddings grounded i
Externí odkaz:
http://arxiv.org/abs/2406.13560
Autor:
Popel, Martin, Poláková, Lucie, Novák, Michal, Helcl, Jindřich, Libovický, Jindřich, Straňák, Pavel, Krabač, Tomáš, Hlaváčová, Jaroslava, Anisimova, Mariia, Chlaňová, Tereza
We present Charles Translator, a machine translation system between Ukrainian and Czech, developed as part of a society-wide effort to mitigate the impact of the Russian-Ukrainian war on individuals and society. The system was developed in the spring
Externí odkaz:
http://arxiv.org/abs/2404.06964
Cross-lingual alignment, the meaningful similarity of representations across languages in multilingual language models, has been an active field of research in recent years. We survey the literature of techniques to improve cross-lingual alignment, p
Externí odkaz:
http://arxiv.org/abs/2404.06228
Autor:
Ali, Adnan Al, Libovický, Jindřich
Neural language models, which reach state-of-the-art results on most natural language processing tasks, are trained on large text corpora that inevitably contain value-burdened content and often capture undesirable biases, which the models reflect. T
Externí odkaz:
http://arxiv.org/abs/2403.13514
Current multimodal models leveraging contrastive learning often face limitations in developing fine-grained conceptual understanding. This is due to random negative samples during pretraining, causing almost exclusively very dissimilar concepts to be
Externí odkaz:
http://arxiv.org/abs/2403.02875
Autor:
Friedrich, Felix, Hämmerl, Katharina, Schramowski, Patrick, Brack, Manuel, Libovicky, Jindrich, Kersting, Kristian, Fraser, Alexander
Text-to-image generation models have recently achieved astonishing results in image quality, flexibility, and text alignment, and are consequently employed in a fast-growing number of applications. Through improvements in multilingual abilities, a la
Externí odkaz:
http://arxiv.org/abs/2401.16092
Autor:
Helcl, Jindřich, Libovický, Jindřich
We present the Charles University system for the MRL~2023 Shared Task on Multi-lingual Multi-task Information Retrieval. The goal of the shared task was to develop systems for named entity recognition and question answering in several under-represent
Externí odkaz:
http://arxiv.org/abs/2310.16528
Autor:
Kydlíček, Hynek, Libovický, Jindřich
Pre-trained models for Czech Natural Language Processing are often evaluated on purely linguistic tasks (POS tagging, parsing, NER) and relatively simple classification tasks such as sentiment classification or article classification from a single ne
Externí odkaz:
http://arxiv.org/abs/2307.10666
Previous work has shown that the representations output by contextual language models are more anisotropic than static type embeddings, and typically display outlier dimensions. This seems to be true for both monolingual and multilingual models, alth
Externí odkaz:
http://arxiv.org/abs/2306.00458