# Tensorboard HParams Granularity

**URL:** <https://discuss.pytorch.org/t/tensorboard-hparams-granularity/58462>\
**Category:** Uncategorized\
**Created:** [October 17, 2019, 8:51am UTC](https://discuss.pytorch.org/t/tensorboard-hparams-granularity/58462 "2019-10-17T08:51:39Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![stevejpurves](https://discuss.pytorch.org/user_avatar/discuss.pytorch.org/stevejpurves/32/9628_2.png) [@stevejpurves](https://discuss.pytorch.org/u/stevejpurves)\
**Post date:** [October 17, 2019, 8:51am UTC](https://discuss.pytorch.org/t/tensorboard-hparams-granularity/58462/1 "2019-10-17T08:51:39Z")

</div>

Hi Folks!

I am using the `add_hparams` function on the integrated Tensorboard SummaryWriter, which is great.  
I notice though that in tensorboard that we do not get any control over the hyperparameter values in tensorboard itself. e.g. compare this image

![image](https://discuss.pytorch.org/uploads/default/original/3X/a/c/ac34bed7531a8574208e43691624913f2cfc6732.png)

to the screenshot in teh tensorboard docs:

> **[Hyperparameter Tuning with the HParams Dashboard  |  TensorBoard...](https://www.tensorflow.org/tensorboard/hyperparameter_tuning_with_hparams)**

I am assuming that this is due to the simpler implementation here (`add_hparams` just receives a dict of params when metrics are logged) where there is no prior declaration of the hyperparameter ranges etc that is available when using the hparams and TF direct.

My question is is there any way currently to register those hparams in tensorboard via the pytorch API or otherwise get the same behavior?  
or any plans to extend in that direction? ( I see no open issues on github)

cheers!