Sebastian Ruder
Sebastian Ruder is a Meta research scientist in Berlin, working on multilingual language models and evaluation. He earlier led Cohere’s multilingual team and researched at Google DeepMind. He studied computational linguistics and completed a doctorate while working at the text-analytics company AYLIEN. Ruder also writes technical explanations and NLP News. His 2016 gradient-descent overview surveyed methods for training models, and in 2018 he coauthored ULMFiT with Jeremy Howard, exploring transfer learning for text classification. He was among the authors of the PaLM 2 report in 2023. In a 2024 essay, he described his move to Cohere.
Timeline
-
Study and doctorate
Language and computation
Ruder studies computational linguistics and completes NLP research alongside work at AYLIEN.
-
2016
Training methods explained
He publishes an overview of gradient-descent algorithms.
-
2018
ULMFiT
Ruder and Jeremy Howard publish a shared transfer-learning method for text classification.
-
2023
PaLM 2
He contributes to the large team report on multilingual and other model capabilities.
-
2024
Cohere and public analysis
Ruder joined Cohere and continues research analysis through NLP News.
-
October 2026
Meta in Berlin
His current research areas include multilingual capabilities and evaluation.
Sources and credits
Photo: Sebastian Ruder personal site. License: Not stated by source. Source: ruder.io. Cropped to a square and resized. Ask us to remove or replace this photo.