doc/RESULTS.md contains "Validation entropy" and "Test entropy" for a range of experiments on the "One Billion Word Benchmark"
The "Validation entropy" number is exactly the same as the "Test entropy" number, so it looks as though the same dataset was used to compute both (they should be news.en.heldout-00000-of-00050 and news.en.heldout-00001-of-00050 respectfully).
doc/RESULTS.md contains "Validation entropy" and "Test entropy" for a range of experiments on the "One Billion Word Benchmark"
The "Validation entropy" number is exactly the same as the "Test entropy" number, so it looks as though the same dataset was used to compute both (they should be news.en.heldout-00000-of-00050 and news.en.heldout-00001-of-00050 respectfully).