large.en model available?

[bentrevett]

From the paper, as far as I'm aware, each model size was trained on both multilingual and English only data. Does this mean that a large.en model was trained, and is there a possibility that it can be released?

Asking as I've found the transcriptions from the medium.en model outperform the large model on English data that I've tested, and would also like to experiment with a large.en model.