`rouge` not working for non-Latin scripts

`rouge` scores for non-Latin scripts are often 0. Is there an option to pass in a custom tokenizer as `rouge-score` does?

```
In [1]: rouge = evaluate.load('rouge')

In [3]: p = r = ['पूर्व प्रधानमन्त्री शिंजो आबेको हत्याले जापान स्तब्ध छ।']

In [4]: rouge.compute(predictions=p, references=r, use_stemmer=False)
Out[4]: {'rouge1': 0.0, 'rouge2': 0.0, 'rougeL': 0.0, 'rougeLsum': 0.0}
```