metadata
language:
- en
tags:
- sentence-simplification
metrics:
- sari
- bleu
pipeline_tag: text2text-generation
Eval results on WikiLarge
We obtain the following results on validation and test sets of WikiLarge:
| Set | SARI | BLEU |
|---|---|---|
| validation | 44.4 | 27.38 |
| test | 39.16 | 31.62 |
EASSE evaluation
TurkCorpus
We obtain the following results on validation and test sets of the Turk corpus:
| Set | SARI | BLEU | FKGL |
|---|---|---|---|
| validation | 38.57 | 87.7 | 8.1 |
| test | 38.82 | 87.89 | 8.49 |
ASSET
We obtain the following results on validation and test sets of the ASSET corpus:
| Set | SARI | BLEU | FKGL |
|---|---|---|---|
| validation | 36.83 | 88.54 | 8.15 |
| test | 35.74 | 86.96 | 8.46 |