Omitted s'

I've only done a simple test but if you remove the whole line 65 from file `tokenizer.py` then the 's' at the end of was and his don't get stripped as it were. This line below:
`text = text.strip("<s>").strip("</s>")`


