arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2007.12770cs.AIcs.CLcs.LG
收藏

BabyAI 1.1

BabyAI 1.1

Mila, Université de Montréal(蒙特利尔大学 Mila 研究所);Element AI(Element AI 公司);CIFAR Senior Fellow(CIFAR 高级研究员)

David Yu-Tung Hui, Maxime Chevalier-Boisvert, Dzmitry Bahdanau, Yoshua Bengio

摘要:

BabyAI平台旨在测量训练代理遵循嵌入式语言指令的样本效率。BabyAI 1.0展示了通过深度模仿或强化学习训练的智能体的基线结果。BabyAI 1.1在三个方面改进了代理的架构。这将强化学习样本效率提高了3倍,并将最难级别上的模仿学习性能从77%提高到90. 4%。我们希望这些改进能够提高BabyAI实验的计算效率,并帮助用户设计更好的代理。

英文摘要:

The BabyAI platform is designed to measure the sample efficiency of training an agent to follow grounded-language instructions. BabyAI 1.0 presents baseline results of an agent trained by deep imitation or reinforcement learning. BabyAI 1.1 improves the agent's architecture in three minor ways. This increases reinforcement learning sample efficiency by up to 3 times and improves imitation learning performance on the hardest level from 77 % to 90.4 %. We hope that these improvements increase the computational efficiency of BabyAI experiments and help users design better agents.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑