Language Modeling on dooley style corpus
40.7gPPLngram model-based logit scaling
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ngram model-based logit scalingweights=00332025.03 | 40.7 | 89.68 | |
| ngram model-based logit scalingweights=00212025.03 | 40.93 | 78.64 | |
| ngram model-based logit scalingweights=00322025.03 | 42.64 | 89.72 | |
| ngram model-based logit scalingweights=11112025.03 | 117.32 | 229.26 | |
| ngram model-based logit scalingweights=11102025.03 | 119.28 | 229.38 | |
| ngram model-based logit scalingweights=01102025.03 | 124.31 | 229.38 | |
| ngram model-based logit scalingweights=01112025.03 | 125.46 | 229.26 |