# How to compare different parameter sets using the validation loss?

**URL:** https://discourse.scverse.org/t/how-to-compare-different-parameter-sets-using-the-validation-loss/2017
**Category:** Help
**Tags:** integration, scvi
**Created:** [January 23, 2024, 1:40pm UTC](https://discourse.scverse.org/t/how-to-compare-different-parameter-sets-using-the-validation-loss/2017 "2024-01-23T13:40:24Z")
**Posts on this page:** 1
**Showing post:** 5

<div class="post-metadata">

### Author: ![swc](https://avatars.discourse-cdn.com/v4/letter/s/54ee81/32.png) [@swc](https://discourse.scverse.org/u/swc)
#### Post date: [August 14, 2024, 4:22pm UTC](https://discourse.scverse.org/t/how-to-compare-different-parameter-sets-using-the-validation-loss/2017/5 "2024-08-14T16:22:04Z")

</div>

Hi Yiwei,

I saw the exactly same thing on my dataset, did you find any explanation for this? I saw the other post mentioning:

> [@Suggestion on parameters for training scvi model](https://discourse.scverse.org/t/suggestion-on-parameters-for-training-scvi-model/1936/2):
>
> If you find that `elbo_validation`, `validation_loss`, and `reconstruction_loss_validation` are converging at the end of training, then your model is probably fine.

Best,

swc

---

_[View the full topic](https://discourse.scverse.org/t/how-to-compare-different-parameter-sets-using-the-validation-loss/2017)._