Výsledky vyhledávání

nmT5 -- Is parallel data still relevant for pre-training massively multilingual language models?

Autor: Aditya Siddhant, Melvin Johnson, Noah Constant, Mihir Kale, Linting Xue, Rami Al-Rfou

Publikováno v: ACL/IJCNLP (2)

Recently, mT5 - a massively multilingual version of T5 - leveraged a unified text-to-text format to attain state-of-the-art results on a wide variety of multilingual NLP tasks. In this paper, we investigate the impact of incorporating parallel data i

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_dedup___::6ecc502f36493adff335cae7874f1a9d
http://arxiv.org/abs/2106.02171

Zobrazit plný text záznamu

Automatic Depression Level Analysis Using Audiovisual Modality

Autor: Mihir Kale, Suhasini Itkar, Mukta Mayee, Preksha Yadav, Aishwarya Chordia

Publikováno v: Smart Computing Techniques and Applications ISBN: 9789811615016

Depression is a mental disorder which is quite common in today’s modern world. Depression has a severe impact on a person’s mood, feelings and health. For early detection and recognition of depression, use of the computer-vision field is found to

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_________::d00430c0ad85c5538488490b7d7061e5
https://doi.org/10.1007/978-981-16-1502-3_43

Zobrazit plný text záznamu

Automatic Construction of Evaluation Suites for Natural Language Generation Datasets

Autor: Simon Mille, Kaustubh Dhole, Saad Mahamood, Laura Perez-Beltrachini, Varun Prashant Gangal, Mihir Kale, Emiel van Miltenburg, Sebastian Gehrmann

Publikováno v: Proceedings of the Neural Information Processing Systems Track on Datasets and Benchmarks 1 (NeurIPS Datasets and Benchmarks 2021)
Tilburg University-PURE

Machine learning approaches applied to NLP are often evaluated by summarizing their performance in a single number, for example accuracy. Since most test sets are constructed as an i.i.d. sample from the overall data, this approach overly simplifies

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_dedup___::8070dedc1ce469faf06434bf5532b1aa

Zobrazit plný text záznamu

The GEM Benchmark: Natural Language Generation, its Evaluation and Metrics

Autor: Thibault Sellam, Vikas Raunak, Cristina Garbacea, Simon Mille, Tatsunori Hashimoto, Miruna-Adriana Clinciu, Ankur P. Parikh, Angelina McMillan-Major, Samira Shaikh, Kaustubh Dhole, Pawan Sasanka Ammanamanchi, Anuoluwapo Aremu, Laura Perez-Beltrachini, Marco Antonio Sobrevilla Cabezudo, Juan Diego Rodriguez, Sashank Santhanam, Bodhisattwa Prasad Majumder, Dhruv Kumar, Esin Durmus, Dipanjan Das, Wei Xu, Wanyu Du, Vitaly Nikolaev, Faisal Ladhak, Moin Nadeem, Chris Chinenye Emezue, Varun Gangal, Diyi Yang, Yangfeng Ji, João Sedoc, Aman Madaan, Shailza Jolly, Saad Mahamood, Karmanya Aggarwal, Emiel van Miltenburg, Antoine Bosselut, Salomey Osei, Jiawei Zhou, Yufang Hou, Harsh Jhamtani, Niranjan Ramesh Rao, Anastasia Shimorina, Shashi Narayan, Akhila Yerukola, Nishant Subramani, Sebastian Gehrmann, Hendrik Strobelt, Khyathi Raghavi Chandu, Yacine Jernite, Tosin P. Adewumi, Mihir Kale, Khyati Mahajan, Andre Niyongabo Rubungo, Pedro Henrique Martins, Ondřej Dušek, Mounica Maddela

Publikováno v: Proceedings of the 1st Workshop on Natural Language Generation, Evaluation, and Metrics (GEM 2021)
Proceedings of the 1st Workshop on Natural Language Generation, Evaluation, and Metrics (GEM 2021), Aug 2021, Online, France. pp.96-120, ⟨10.18653/v1/2021.gem-1.10⟩
Tilburg University-PURE
Proceedings of the 1st Workshop on Natural Language Generation, Evaluation, and Metrics (GEM 2021), 96-120
STARTPAGE=96;ENDPAGE=120;TITLE=Proceedings of the 1st Workshop on Natural Language Generation, Evaluation, and Metrics (GEM 2021)

International audience; We introduce GEM, a living benchmark for natural language Generation (NLG), its Evaluation, and Metrics. Measuring progress in NLG relies on a constantly evolving ecosystem of automated metrics, datasets, and human evaluation

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_dedup___::56549f809a95ad121bc57aac47f790d8
https://doi.org/10.18653/v1/2021.gem-1.10

Zobrazit plný text záznamu

mT5: A Massively Multilingual Pre-trained Text-to-Text Transformer

Autor: Noah Constant, Aditya Siddhant, Aditya Barua, Linting Xue, Mihir Kale, Adam Roberts, Rami Al-Rfou, Colin Raffel

Publikováno v: NAACL-HLT

The recent "Text-to-Text Transfer Transformer" (T5) leveraged a unified text-to-text format and scale to attain state-of-the-art results on a wide variety of English-language NLP tasks. In this paper, we introduce mT5, a multilingual variant of T5 th

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_dedup___::6841f14537777a14674607884ca00048
https://doi.org/10.18653/v1/2021.naacl-main.41

Zobrazit plný text záznamu

MixOut: A Simple Yet Effective Data Augmentation Scheme for Slot-Filling

Autor: Aditya Siddhant, Mihir Kale

Publikováno v: Lecture Notes in Electrical Engineering ISBN: 9789811583940
IWSDS

We present a data augmentation strategy for slot-filling in task-oriented dialogue systems. It is simple yet effective and does not rely on external corpora. Lexicons for all slot types are generated from available annotated data. Synthetic, yet real

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_________::aee3811ff6a30f54dd68fdee99db42ed
https://doi.org/10.1007/978-981-15-8395-7_21

Zobrazit plný text záznamu

Template Guided Text Generation for Task-Oriented Dialogue

Autor: Mihir Kale, Abhinav Rastogi

Publikováno v: EMNLP (1)

Virtual assistants such as Google Assistant, Amazon Alexa, and Apple Siri enable users to interact with a large number of services and APIs on the web using natural language. In this work, we investigate two methods for Natural Language Generation (N

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_dedup___::b7ecd11e060df083bc031e91e21fb21e
http://arxiv.org/abs/2004.15006

Zobrazit plný text záznamu

TicketTalk: Toward human-level performance with end-to-end, transaction-based dialog systems

Autor: Bill Byrne, Mihir Kale, Karthik Krishnamoorthi, Saravanan Ganesh

Publikováno v: ACL/IJCNLP (1)

We present a data-driven, end-to-end approach to transaction-based dialog systems that performs at near-human levels in terms of verbal response quality and factual grounding accuracy. We show that two essential components of the system produce these

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_dedup___::2f2c57bea87f3cdb9df65b223cf13fb2

Zobrazit plný text záznamu

Vyhledávací nástroje:

Upřesnit hledání