Text Generation on Open Web Text (OWT) (val)
0.456GPT-2 GM ScoreMasked D-MMD
Evaluation Results
| Method | Links | |
|---|---|---|
| Masked D-MMDp=0.85, NFE=82026.03 | 0.456 | |
| Masked Teacherp=0.85, NFE=322026.03 | 0.402 | |
| Uniform Teacherp=0.50, NFE=322026.03 | 0.375 | |
| Uniform D-MMDp=0.70/0.70, NFE=82026.03 | 0.337 | |
| Uniform Teacherp=0.50, NFE=1282026.03 | 0.33 | |
| Uniform Teacherp=0.50, NFE=642026.03 | 0.326 | |
| Uniform Teacherp=0.50, NFE=2562026.03 | 0.324 | |
| Uniform D-MMDp=0.70/0.70, NFE=642026.03 | 0.316 | |
| Uniform Teacherp=0.50, NFE=5122026.03 | 0.313 | |
| Uniform D-MMDp=0.70/0.70, NFE=162026.03 | 0.31 | |
| Uniform D-MMDp=0.70/0.70, NFE=322026.03 | 0.307 | |
| Masked Teacherp=0.85, NFE=642026.03 | 0.307 | |
| Masked Teacherp=0.85, NFE=1282026.03 | 0.297 | |
| Masked Teacherp=0.85, NFE=2562026.03 | 0.275 | |
| Masked Teacherp=0.85, NFE=5122026.03 | 0.275 | |
| Masked D-MMDp=0.85, NFE=162026.03 | 0.236 | |
| Masked D-MMDp=0.85, NFE=642026.03 | 0.231 | |
| Masked D-MMDp=0.85, NFE=322026.03 | 0.225 | |
| AR BaselineNFE=642026.03 | 0.061 |