Language Modeling on WikiText-103 (Mauve and ROUGE Metrics)
87MauveMLE (ours)
Evaluation results
| Method | Links | ||
|---|---|---|---|
| MLE (ours)Model=OPT-125M2026.05 | 87 | 41 | |
| MLE (ours)Model=GPT-22026.05 | 87 | 40 | |
| MLE (ours)Model=GPT-Neo-1.3B2026.05 | 87 | 41 | |
| EMOModel=GPT-Neo-1.3B2026.05 | 85 | 38 | |
| EMOModel=GPT-22026.05 | 82 | 39 | |
| MLEModel=GPT-Neo-1.3B2026.05 | 82 | 39 | |
| TVDModel=GPT-Neo-1.3B2026.05 | 81 | 39 | |
| MLEModel=OPT-125M2026.05 | 80 | 38 | |
| MixCEModel=OPT-125M2026.05 | 79 | 38 | |
| EMOModel=OPT-125M2026.05 | 79 | 38 | |
| MixCEModel=GPT-Neo-1.3B2026.05 | 79 | 39 | |
| BrierModel=GPT-Neo-1.3B2026.05 | 79 | 38 | |
| TVDModel=OPT-125M2026.05 | 77 | 38 | |
| MixCEModel=GPT-22026.05 | 77 | 38 | |
| TVDModel=GPT-22026.05 | 75 | 37 | |
| MLEModel=GPT-22026.05 | 74 | 37 | |
| BrierModel=GPT-22026.05 | 74 | 36 | |
| BrierModel=OPT-125M2026.05 | 67 | 37 |