Fetching the paper…

Eval all, trust a few, do wrong to none: Comparing sentence generation models · Around