PixL2R: Guiding Reinforcement Learning Using Natural Language by Mapping Pixels to Rewards

Open in new window