Timezone: »

 
Poster
StreamingQA: A Benchmark for Adaptation to New Knowledge over Time in Question Answering Models
Adam Liska · Tomas Kocisky · Elena Gribovskaya · Tayfun Terzi · Eren Sezener · Devang Agrawal · Cyprien de Masson d'Autume · Tim Scholtes · Manzil Zaheer · Susannah Young · Ellen Gilsenan-McMahon · Sophia Austin · Phil Blunsom · Angeliki Lazaridou

Tue Jul 19 03:30 PM -- 05:30 PM (PDT) @ Hall E #128

Knowledge and language understanding of models evaluated through question answering (QA) has been usually studied on static snapshots of knowledge, like Wikipedia, however, our world is dynamic, evolves over time, and our models’ knowledge becomes outdated. To study how semi-parametric QA models and their underlying parametric language models (LMs) adapt to evolving knowledge, we construct a new large-scale dataset, StreamingQA, with human written and generated questions asked on a given date, to be answered from 14 years of timestamped news articles. We evaluate our models quarterly as they read new articles not seen in pre-training. We show that parametric models can be updated without full retraining, while avoiding catastrophic forgetting. For semi-parametric models, adding new articles into the search space allows for rapid adaptation, however, models with an outdated underlying LM under-perform those with a retrained LM. For questions about higher-frequency named entities, parametric updates are particularly beneficial. In our dynamic world, the StreamingQA dataset enables a more realistic evaluation of QA models, and our experiments highlight several promising directions for future research.

Author Information

Adam Liska (DeepMind)
Tomas Kocisky (DeepMind)
Elena Gribovskaya (DeepMind)
Tayfun Terzi (DeepMind)
Eren Sezener (DeepMind)
Devang Agrawal (DeepMind)
Cyprien de Masson d'Autume (DeepMind)
Tim Scholtes (DeepMind)
Manzil Zaheer (Google Research)
Susannah Young (DeepMind)
Ellen Gilsenan-McMahon (DeepMind)
Sophia Austin (DeepMind)
Phil Blunsom (DeepMind and Oxford University)
Angeliki Lazaridou (DeepMind)

Related Events (a corresponding poster, oral, or spotlight)

More from the Same Authors