Created by: ksellesk

We can not get a WER 6.85% with CommonCrawl(en.00) Language model.

Do you use the CommonCrawl(en.00) Language model in the LibriSpeech benchmark?