No it does not. In the past we felt that the best way to achieve our goals was to focus almost exclusively on training large models though, and we no longer feel that’s the case.