# Average multiple runs over one parameter

**URL:** <https://community.wandb.ai/t/average-multiple-runs-over-one-parameter/4371>\
**Category:** W&B Help\
**Tags:** sweeps, wandb\
**Created:** [May 10, 2023, 7:29pm UTC](https://community.wandb.ai/t/average-multiple-runs-over-one-parameter/4371 "2023-05-10T19:29:05Z")\
**Posts on this page:** 5\
**Page:** 1

<div class="post-metadata">

**Author:** ![parisi](https://avatars.discourse-cdn.com/v4/letter/p/35a633/32.png) [@parisi](https://community.wandb.ai/u/parisi)\
**Post date:** [May 10, 2023, 7:29pm UTC](https://community.wandb.ai/t/average-multiple-runs-over-one-parameter/4371/1 "2023-05-10T19:29:05Z")

</div>

I am doing a sweep over multiple parameters. For every configuration, I also have a seed to control randomness.  
For instance, let’s say I am sweeping over learning rate in [0.1, 0.01] and batch size in [16, 32]. This will give me 4 runs, and for each I will have two runs with seed 1 and 2, treated as hyperparameter, for a total of 8 runs.  
I’d like to average ALL plots over the 2 seeds. So, instead of having 8 curves per plot I’ll have only 4.  
I can do this manually by grouping over learning rate and batch size. However, in practice, I have way more hyperparameters than 2, and I don’t want to manually add all of them one by one to the group.

Is there a way to, basically, “group over all but one” hyperparameter?

---

<div class="post-metadata">

**Author:** ![raphael-sanandres](https://avatars.discourse-cdn.com/v4/letter/r/7cd45c/32.png) [@raphael-sanandres](https://community.wandb.ai/u/raphael-sanandres)\
**Post date:** [May 11, 2023, 5:05pm UTC](https://community.wandb.ai/t/average-multiple-runs-over-one-parameter/4371/2 "2023-05-11T17:05:02Z")

</div>

Hello @parisi !

Thank you for bringing this up! We do not support this currently but I can make a feature request for you for our internal team to look at.

---

<div class="post-metadata">

**Author:** ![parisi](https://avatars.discourse-cdn.com/v4/letter/p/35a633/32.png) [@parisi](https://community.wandb.ai/u/parisi)\
**Post date:** [May 11, 2023, 5:17pm UTC](https://community.wandb.ai/t/average-multiple-runs-over-one-parameter/4371/3 "2023-05-11T17:17:19Z")

</div>

That would be great, thanks!

---

<div class="post-metadata">

**Author:** ![jlperla](https://sea2.discourse-cdn.com/flex020/user_avatar/community.wandb.ai/jlperla/32/963_2.png) [@jlperla](https://community.wandb.ai/u/jlperla)\
**Post date:** [May 15, 2023, 5:24pm UTC](https://community.wandb.ai/t/average-multiple-runs-over-one-parameter/4371/4 "2023-05-15T17:24:32Z")

</div>

Yes, I do this as well and more tooling would help.

More broadly, when doing a HPO it would be great to be able to tell it to run a set number of seeds at each hyperparameter value and then use the median (or whatever) as the HPO objective value.

---

<div class="post-metadata">

**Author:** ![system](https://us1.discourse-cdn.com/flex020/uploads/wandb/original/1X/366b649231631dbab896843020da0056074ac79d.png) [@system](https://community.wandb.ai/u/system)\
**Post date:** [July 14, 2023, 5:25pm UTC](https://community.wandb.ai/t/average-multiple-runs-over-one-parameter/4371/5 "2023-07-14T17:25:02Z")

</div>

Closed on July 14, 2023, 5:25pm UTC

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.