GitLab

* Add attention masking to attn processors

* Update tensor conversion

---------
Co-authored-by: YiYi Xu <yixu310@gmail.com>
Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>