BabyAI 1.1
This is an incremental improvement for researchers using the BabyAI platform to design more computationally efficient agents.
The paper tackles improving sample efficiency for training agents to follow grounded-language instructions on the BabyAI platform, resulting in up to 3 times better reinforcement learning efficiency and imitation learning performance increasing from 77% to 90.4% on the hardest level.
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.