Alchemy

DeepMind Alchemy 环境是一个元强化学习基准，它呈现从具有深层底层结构的任务分布中采样的任务。它的创建是为了测试代理通过潜在状态推理以及有用的探索和实验进行推理和计划的能力。 Alchemy 是一款单人视频游戏，在 Unity 中实现。玩家看到一张桌子的第一人称视图，上面有许多物品，包括一组彩色宝石、一组装有彩色药水的盘子和一个中央大锅。石头有不同的点值，将石头添加到大锅中时会收集点数。通过将石头浸入药水中，玩家可以改变石头的外观，从而改变它们的价值，增加可以赢得的点数。The DeepMind Alchemy environment is a meta-reinforcement learning benchmark that presents tasks sampled from a task distribution with a deep underlying structure. It was developed to test agents' ability to reason and plan through latent state reasoning, as well as productive exploration and experimentation. Alchemy is a single-player video game implemented in the Unity game engine. Players are presented with a first-person view of a table with various items, including a set of colored gemstones, a set of trays holding colored potions, and a central cauldron. Each stone has a distinct point value, and points are collected when stones are added to the cauldron. By submerging stones in the potions, players can modify the stones' appearances, which in turn changes their point values and increases the total points that can be won.

搜集汇总

背景与挑战

背景概述

Alchemy是一个由DeepMind开发的元强化学习基准环境，以Unity实现的单人视频游戏形式呈现，玩家通过将宝石浸入药水来改变其外观和价值，以测试智能体的推理和探索能力。