Search Results - "Ring, Roman"
-
1
Grandmaster level in StarCraft II using multi-agent reinforcement learning
Published in Nature (London) (01-11-2019)“…Many real-world applications require artificial agents to compete and coordinate with other agents in complex environments. As a stepping stone to this goal,…”
Get full text
Journal Article -
2
Red Teaming Language Models with Language Models
Published 07-02-2022“…Language Models (LMs) often cannot be deployed because of their potential to harm users in hard-to-predict ways. Prior work identifies harmful behaviors before…”
Get full text
Journal Article -
3
AlphaStar Unplugged: Large-Scale Offline Reinforcement Learning
Published 07-08-2023“…StarCraft II is one of the most challenging simulated reinforcement learning environments; it is partially observable, stochastic, multi-agent, and mastering…”
Get full text
Journal Article -
4
Flamingo: a Visual Language Model for Few-Shot Learning
Published 29-04-2022“…Building models that can be rapidly adapted to novel tasks using only a handful of annotated examples is an open challenge for multimodal machine learning…”
Get full text
Journal Article -
5
Improving language models by retrieving from trillions of tokens
Published 08-12-2021“…We enhance auto-regressive language models by conditioning on document chunks retrieved from a large corpus, based on local similarity with preceding tokens…”
Get full text
Journal Article -
6
Scaling Language Models: Methods, Analysis & Insights from Training Gopher
Published 08-12-2021“…Language modelling provides a step towards intelligent communication systems by harnessing large repositories of written human knowledge to better predict and…”
Get full text
Journal Article