Language Modeling on 100 Billion Word Google News Dataset (test)
38.2Test Perplexity (0.1 epochs)
MoE-16384-h
Updated 4d ago
Evaluation Results
| Method | Links | ||
|---|---|---|---|
2017.01 | 38.2 | 29.7 | |
2017.01 | 38.2 | 28.9 | |
2017.01 | 38.9 | 30.9 | |
2017.01 | 39.8 | 29.2 | |
2017.01 | 40.3 | 32.7 | |
2017.01 | 42.8 | 35.3 | |
2017.01 | 48.5 | 40.4 | |
2017.01 | 54.5 | 47 | |
2017.01 | 67.1 | 45.3 |