Bridging the gap to real-world language-grounded visual concept learning | Read Paper on Bytez