# NN Architecture

**URL:** <https://www.alphagenomecommunity.com/t/nn-architecture/891>\
**Category:** Feedback & Feature Requests\
**Created:** [May 11, 2026, 7:25pm UTC](https://www.alphagenomecommunity.com/t/nn-architecture/891 "2026-05-11T19:25:14Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![Christian\_Frey](https://yyz2.discourse-cdn.com/watermelon/user_avatar/www.alphagenomecommunity.com/christian_frey/32/337_2.png) [@Christian\_Frey](https://www.alphagenomecommunity.com/u/Christian_Frey)\
**Post date:** [May 11, 2026, 7:25pm UTC](https://www.alphagenomecommunity.com/t/nn-architecture/891/1 "2026-05-11T19:25:14Z")

</div>

I think, given the long-sequence dynamics of DNA, it would be worthwhile to try the SigFormer Architecture (instead of a pure Transformer architecture), combining Mamba and Transformer Architecture and keeping the computational costs manageable.
