Search for a command to run...
Towards Learning a Generic Agent for Vision-and-Language Navigation via Pre-training