Benjamin Warner, Antoine Chaffin, Benjamin Clavié, Orion Weller, Oskar Hallström, Said Taghadouini, Alexis Gallagher, Raja Biswas, Faisal Ladhak, Tom Aarsen, Griffin Thomas Adams, Jeremy Howard, Iacopo Poli. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Julian Eisenschlos, Sebastian Ruder, Piotr Czapla, Marcin Kadras, Sylvain Gugger, Jeremy Howard. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP). 2019.

Inductive transfer learning has greatly impacted computer vision, but existing approaches in NLP still require task-specific modifications and training from scratch.We propose Universal Language Model Fine-tuning (ULMFiT), an effective transfer learning method that can be applied to any task in NLP, and introduce techniques that are key for fine-tuning a language model.Our method significantly outperforms the state-of-the-art on six text classification tasks, reducing the error by 18-24% on the majority of datasets.Furthermore, with only 100 labeled examples, it matches the performance of training from scratch on 100× more data.We opensource our pretrained models and code 1 .

It's not often we get to peer into the details of how big corporates use predictive modeling in practice. In this talk, Sprint's Head of Predictive Modeling, Tracey De Poalo, will talk about the process she developed using SAS and logistic regression to build a wide range of models. Jeremy Howard will discuss his experience as a consultant at Sprint comparing R and random forests to the existing process, and will show the pros and cons of each approach.

In the last year deep learning has gone from being a special purpose machine learning technique used mainly for image and speech recognition, to becoming a general purpose machine learning tool. This has broad implications for all organizations that rely on data analysis. It represents the latest development in a general trend towards more automated algorithms, and away from domain specific knowledge. For organizations that rely on domain expertise for their competitive advantage, this trend could be extremely disruptive. For start-ups interested in entering established markets, this trend could be a major opportunity. This talk will be a non-technical introduction to general-purpose deep learning, and its potential business impact.