1 Billion Word Language Model Benchmark R13 Output is a new benchmark corpus used to measure and count progress in language modeling. With nearly 1 billion words of training data, the benchmark can quickly evaluate new language modeling techniques and combine them with other new technologies.

This dataset was released by Cornell University in 2013, and the main publishers are Ciprian Chelba, Tomas Mikolov, Mike Schuster, Qi Ge, Thorsten Brants, Phillipp Koehn and Tony Robinson.