Sebastian Ruder

JUNE 6, 2022

ACL 2022 took place in Dublin from 22nd–27th May 2022. Language diversity and multimodality Panelists and their spoken languages at the ACL 2022 keynote panel on supporting linguistic diversity. We can also incorporate additional knowledge by modifying the training data, e.g., by inserting metadata strings (e.g.,

NLP 52

NLP Natural Language Processing Computational Linguistics Neural Network 52

Marek Rei

NOVEMBER 4, 2024

EMNLP 2022. EMNLP 2022. NeurIPS 2022. EMNLP 2022. EMNLP 2022. They show performance improvements in some settings and speed improvements in all evaluated settings, showing particular usefulness in settings where the LLM needs to retrieve information about multiple entities (e.g. UC Berkeley, CMU. Google Research.

Machine Learning 80

Machine Learning NLP Large Language Models LLM 80

Topbots

JUNE 15, 2023

70% of research papers published in a computational linguistics conference only evaluated English.[ I additionally use metadata from The World Atlas of Language Structures to obtain information such as language family (e.g. Association for Computational Linguistics. Association for Computational Linguistics.

Natural Language Processing 59

Natural Language Processing Computational Linguistics NLP ChatGPT 59

Sebastian Ruder

NOVEMBER 14, 2022

This post is partially based on a keynote I gave at the Deep Learning Indaba 2022. Linguistic and demographic global utility metrics for a number of language technology tasks. Overall, however, we observe very low linguistic utility numbers, showing an unequal performance distribution across the world's languages.

Natural Language Processing 52

Natural Language Processing NLP Computational Linguistics BERT 52

Sebastian Ruder

FEBRUARY 23, 2023

For modular fine-tuning for NLP, check out our EMNLP 2022 tutorial. d) Hypernetwork: A small separate neural network generates modular parameters conditioned on metadata. We provide a high-level overview of some of the trade-offs of the different computation functions below. For a more in-depth review, refer to our survey.

Deep Learning 52

Deep Learning NLP Computer Vision Metadata 52