# Adding additional bands

**URL:** <https://community.drivendata.org/t/adding-additional-bands/7022>\
**Category:** On Cloud N\
**Created:** [January 7, 2022, 12:01pm UTC](https://community.drivendata.org/t/adding-additional-bands/7022 "2022-01-07T12:01:32Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![ajijohn](https://avatars.discourse-cdn.com/v4/letter/a/5f9b8f/32.png) [@ajijohn](https://community.drivendata.org/u/ajijohn)\
**Post date:** [January 7, 2022, 12:01pm UTC](https://community.drivendata.org/t/adding-additional-bands/7022/1 "2022-01-07T12:01:32Z")

</div>

Hey folks,

I want to try band differences as an additional input (like the Sentinel cloud algorithm), but I get loss as nan.0

[06:57\<00:00, 2.81it/s, loss=nan, v\_num=3, loss\_step=nan.0, iou\_step=0.000, iou\_epoch=0.000, loss\_epoch=nan.0]

I modified the input channels to 5, and did the band difference and appended to the new band array on the fly.

Am I missing anything ? My expectation is that similar to a new band input (for eg SWIR band) that is physically on the disk, this ought to improve.

I’m using the Unet benchmark code given by the organizers.