L
Laurent Sifre
Researcher at Google
Publications - 33
Citations - 35068
Laurent Sifre is an academic researcher from Google. The author has contributed to research in topics: Reinforcement learning & Computer science. The author has an hindex of 15, co-authored 23 publications receiving 22890 citations. Previous affiliations of Laurent Sifre include École Normale Supérieure.
Papers
More filters
Proceedings ArticleDOI
Large-Scale Retrieval for Reinforcement Learning
Peter C. Humphreys,Arthur Guez,Olivier Tieleman,Laurent Sifre,Theophane Weber,Timothy P. Lillicrap +5 more
TL;DR: This work pursues an alternative approach in which agents can utilise large-scale contextsensitive database lookups to support their parametric computations, which allows agents to directly learn in an end-to-end manner to utilise relevant information to inform their outputs.
Posted Content
Machine Translation Decoding beyond Beam Search
Rémi Leblond,Jean-Baptiste Alayrac,Laurent Sifre,Miruna Pislar,Jean-Baptiste Lespiau,Ioannis Antonoglou,Karen Simonyan,Oriol Vinyals +7 more
TL;DR: In this article, Monte-Carlo Tree Search (MCTS) based method was used for decoding auto-regressive machine translation models, and it showed promising results on a variety of metrics.
Proceedings Article
Muesli: Combining Improvements in Policy Optimization
Matteo Hessel,Ivo Danihelka,Fabio Viola,Arthur Guez,Simon Schmitt,Laurent Sifre,Theophane Weber,David Silver,Hado van Hasselt +8 more
TL;DR: Muesli as discussed by the authors combines regularized policy optimization with model learning as an auxiliary loss to match MuZero's state-of-the-art performance on Atari, and has computation speed comparable to model-free baselines.
Cross-lingual and progressive transfer learning
TL;DR: The authors proposed a cross-lingual and progressive transfer learning approach, called CLP-Transfer, that transfers models from a source language, for which pretrained models are available to a new target language.
This is the way - lessons learned from designing and compiling LEPISZCZE, a comprehensive NLP benchmark for Polish
TL;DR: The LEPISZCZE benchmark as mentioned in this paper is a comprehensive benchmark for Polish NLP with a large variety of tasks and high-quality operationalization of the KLEJ benchmark.