Language Modeling on Remus style corpus
51.98gPPLngram model-based logit scaling
Evaluation results
| Method | Links | ||
|---|---|---|---|
| ngram model-based logit scalingweights=00212025.03 | 51.98 | 78.73 | |
| ngram model-based logit scalingweights=00322025.03 | 52.14 | 82.9 | |
| ngram model-based logit scalingweights=00212025.03 | 55.15 | 54.92 | |
| ngram model-based logit scalingweights=00322025.03 | 57.21 | 47.82 | |
| ngram model-based logit scalingweights=00332025.03 | 57.29 | 47.79 | |
| ngram model-based logit scalingweights=00332025.03 | 58.43 | 82.88 | |
| ngram model-based logit scalingweights=11112025.03 | 86.61 | 36.94 | |
| ngram model-based logit scalingweights=01112025.03 | 87.55 | 36.94 | |
| ngram model-based logit scalingweights=11112025.03 | 89.57 | 50.34 | |
| ngram model-based logit scalingweights=01102025.03 | 92.08 | 36.99 | |
| ngram model-based logit scalingweights=01102025.03 | 96.95 | 50.35 | |
| ngram model-based logit scalingweights=11102025.03 | 99.95 | 50.35 | |
| ngram model-based logit scalingweights=11102025.03 | 108.04 | 36.99 | |
| ngram model-based logit scalingweights=01112025.03 | 109.87 | 50.34 |