Abstract
Inspired by the diversity and depth of XLand and the simplicity and minimalism of MiniGrid, we present XLand-MiniGrid, a suite of tools and grid-world environ-ments for meta-reinforcement learning research. Written in JAX, XLand-MiniGrid is designed to be highly scalable and can potentially run on GPU or TPU acceler-
XLand-MiniGrid：JAX中的元强化学习利器
受XLand的多样性和深度以及MiniGrid的简单性和极简主义的启发,我们推出了XLand-MiniGrid,这是一套用于元强化学习研究的工具和网格世界环境。XLand-MiniGrid是用JAX编写的,它被设计成高度可扩展的,并且有可能在GPU或TPU加速器上运行,从而在有限的资源下实现大规模实验的民主化。
XLand-MiniGrid: Scalable Meta-Reinforcement Learning
We present XLand-Minigrid, a suite of tools and grid-world environments for meta-reinforcement learning research inspired by the diversity and depth of XLand and the simplicity and minimalism of MiniGrid. XLand-Minigrid is written in JAX, designed to be highly scalable, and can potentially run on GPU or TPU accelerators, democratizing large-scale
XLand-100B: A Large-Scale Multi-Task Dataset for In-Context
We present XLand-100B, a large-scale dataset for in-context reinforcement learning based on the XLand-MiniGrid environment, as a first step to alleviate this problem. It contains complete learning histories for nearly 30,000 different tasks, covering 100B transitions and 2.5B episodes.
Российские ученые создали платформу для контекстного
XLand-MiniGrid появился, чтобы закрыть этот пробел», — пояснил Вячеслав Синий из T-Bank AI Research. Руководитель группы «Адаптивные агенты» Владислав Куренков добавил, что благодаря разнообразию задач
Paper page
XLand-Minigrid is written in JAX, designed to be highly scalable, and can potentially run on GPU or TPU accelerators, democratizing large-scale experimentation with limited resources. To demonstrate the generality of our library, we have implemented some well-known single-task environments as well as new meta-learning environments capable of
XLand-MiniGrid: Scalable Meta-Reinforcement Learning
Written in JAX, XLand-MiniGrid is designed to be highly scalable and can potentially run on GPU or TPU accelerators, democratizing large-scale experimentation with limited resources. Along with the environments, XLand-MiniGrid provides pre-sampled benchmarks with millions of unique tasks of varying difficulty and easy-to-use baselines that
MiniGrid Documentation
Minigrid contains simple and easily configurable grid world environments to conduct Reinforcement Learning research. This library was previously known as gym-minigrid. Toggle site navigation sidebar. MiniGrid Documentation. Farama Foundation Hide navigation sidebar. Hide table of contents sidebar
xminigrid
XLand-MiniGrid is a suite of tools, grid-world environments and benchmarks for meta-reinforcement learning research inspired by the diversity and depth of XLand and the simplicity and minimalism of MiniGrid. Despite the similarities, XLand-MiniGrid is written in JAX from scratch and designed to be highly scalable, democratizing large-scale
Российские учёные создали первую открытую среду для
Inspired by the diversity and depth of XLand and the simplicity and minimalism of MiniGrid, we present XLand-MiniGrid, a suite of tools and grid-world environments for meta-reinforcement learning research. Written in JAX, XLand-MiniGrid is designed to be highly scalable and can potentially run on GP
Abstract arXiv:2312.12044v2 [cs.LG] 6 Feb 2024
Inspired by the diversity and depth of XLand and the simplicity and minimalism of MiniGrid, we present XLand-MiniGrid, a suite of tools and grid-world environments for meta-reinforcement learn-ing research. Written in JAX, XLand-MiniGrid is designed to be highly scalable and can poten-tially run on GPU or TPU accelerators, democ-
XLand-MiniGrid: Scalable Meta-Reinforcement
Written in JAX, XLand-MiniGrid is designed to be highly scalable and can potentially run on GPU or TPU accelerators, democratizing large-scale experimentation with limited resources. Along with the
GitHub
XLand-MiniGrid is a suite of tools, grid-world environments and benchmarks for meta-reinforcement learning research inspired by the diversity and depth of XLand and the simplicity and minimalism of MiniGrid. Despite the similarities, XLand-MiniGrid is written in JAX from scratch and designed to be highly scalable, democratizing large-scale
XLand-MiniGrid: Scalable Meta-Reinforcement Learning
We present XLand-MiniGrid, a suite of tools and grid-world environments for meta-reinforcement learning research inspired by the diversity and depth of XLand and the simplicity and minimalism of MiniGrid. XLand-Minigrid is written in JAX, designed to be highly scalable, and can potentially run on GPU or TPU accelerators, democratizing large
NeurIPS Poster XLand-MiniGrid: Scalable Meta-Reinforcement
Written in JAX, XLand-MiniGrid is designed to be highly scalable and can potentially run on GPU or TPU accelerators, democratizing large-scale experimentation with limited resources. Along with the environments, XLand-MiniGrid provides pre-sampled benchmarks with millions of unique tasks of varying difficulty and easy-to-use baselines that
XLand-MiniGrid: Scalable Meta-Reinforcement Learning
Written in JAX, XLand-MiniGrid is designed to be highly scalable and can potentially run on GPU or TPU accelerators, democratizing large-scale experimentation with limited resources. Along with the environments, XLand-MiniGrid provides pre-sampled benchmarks with millions of unique tasks of varying difficulty and easy-to-use baselines that
XLand-MiniGrid: Scalable Meta-Reinforcement Learning
We present XLand-MiniGrid, a suite of tools and grid-world environments for meta-reinforcement learning research inspired by the diversity and depth of XLand and the simplicity and minimalism of MiniGrid. XLand-Minigrid is written in JAX, designed to be highly scalable, and can potentially run on GPU or TPU accelerators, democratizing large
Paper page
We present XLand-MiniGrid, a suite of tools and grid-world environments for meta-reinforcement learning research inspired by the diversity and depth of XLand and the simplicity and minimalism of MiniGrid. XLand-Minigrid is written in JAX, designed to be highly scalable, and can potentially run on GPU or TPU accelerators, democratizing large
XLand-MiniGrid: Scalable Meta-Reinforcement Learning
introduce XLand-MiniGrid, a library of grid world environments for meta-RL research. It does not compromise on task complexity in favour of affordability, democratizing large scale experimentation with limited resources. 2 XLand-MiniGrid We present an initial release of XLand-MiniGrid(v0.0.1), a suit of tools and grid world environments
xland-minigrid 开源项目教程
文章浏览阅读413次,点赞5次,收藏3次。xland-minigrid 开源项目教程 xland-minigrid JAX-accelerated Meta-Reinforcement Learning Environments Inspired by XLand and MiniGrid ????️_xland 强化学习
XLand-100B: A Large-Scale Multi-Task Dataset for In
We present XLand-100B, a large-scale dataset for in-context reinforcement learning based on the XLand-MiniGrid environment, as a first step to alleviate this problem. It contains complete learning histories for nearly
[2312.12044] XLand-MiniGrid: Scalable Meta-Reinforcement
Written in JAX, XLand-MiniGrid is designed to be highly scalable and can potentially run on GPU or TPU accelerators, democratizing large-scale experimentation with limited resources. Along with the environments, XLand-MiniGrid provides pre-sampled benchmarks with millions of unique tasks of varying difficulty and easy-to-use baselines that
NeurIPS Poster XLand-MiniGrid: Scalable Meta-Reinforcement
Written in JAX, XLand-MiniGrid is designed to be highly scalable and can potentially run on GPU or TPU accelerators, democratizing large-scale experimentation with limited resources. Along
6 FAQs about [Liechtenstein xland minigrid]
What is xLand-minigrid?
Along with the environments, XLand-MiniGrid provides pre-sampled benchmarks with millions of unique tasks of varying difficulty and easy-to-use baselines that allow users to quickly start training adaptive agents.
Can xLand-minigrid help practitioners perform meta-reinforcement learning experiments faster?
ck time. While we do not introduce any novel algorithmic improvements in our work, we hope that the proposed highly scalable XLand-MiniGrid environments will help practitioners perform meta-reinforcement learning experiments at scale faster and with fewer r
What is xLand-minigrid environment interface?
Similar to Jumanji (Bonnet et al., 2023), XLand-MiniGrid Environment interface is inspired by the dm_env API (Muldal et al., 2019), which is particularly well suited for the meta-RL, as it separates episodes from trials by design (see Section D.1 ). Thus, each environment should provide jit-compatible reset, reset_trial and step methods.
Is xLand-minigrid a asynchronous vectorization?
For single-tasks environments we consider random policy and PPO. As can be seen, compared to the commonly used MiniGrid (Chevalier-Boisvert et al., 2023) environments with gymnasium (Towers et al., 2023) asynchronous vectorization, XLand-Minigrid achieves at least 10x faster throughput reaching tens of millions of steps per second.
How many rules can xLand-minigrid use?
Full-scale XLand environment can use more than five rules according to the Team et al. ( 2023). To test XLand-MiniGrid in similar conditions we report simulation throughput varying number of rules. For testing purposes we just replicated same NEAR rule multiple times in the PutNear environment.
Does xLand-minigrid support recurrent PPO?
aselinesWith the release of XLand-MiniGrid, we are providing near-single-file implementations of recurrent PPO (Schulman et al., 2017) for single-task environments and its extension to RL2 (Duan et al., 2016; Wang et al., 2016) for meta-learning as b