# Training loss getting worse over time but training is not stopped

**URL:** <https://discuss.roboflow.com/t/training-loss-getting-worse-over-time-but-training-is-not-stopped/4925>\
**Category:** 🤝  Community Help\
**Created:** [March 19, 2024, 8:30pm UTC](https://discuss.roboflow.com/t/training-loss-getting-worse-over-time-but-training-is-not-stopped/4925 "2024-03-19T20:30:19Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![Jonas\_Haag](https://yyz1.discourse-cdn.com/flex029/user_avatar/discuss.roboflow.com/jonas_haag/32/2744_2.png) [@Jonas\_Haag](https://discuss.roboflow.com/u/Jonas_Haag)\
**Post date:** [March 19, 2024, 8:30pm UTC](https://discuss.roboflow.com/t/training-loss-getting-worse-over-time-but-training-is-not-stopped/4925/1 "2024-03-19T20:30:19Z")

</div>

The metrics of my trainings get worse over time but training does not stop. If I understand correctly, the final model that is picked by Roboflow is worse than one of the earlier checkpoints.

 ![Screenshot 2024-03-19 at 21.27.13](https://canada1.discourse-cdn.com/flex029/uploads/roboflow1/original/2X/5/5c927d998ce7284f9fc4d97469917ccb76f24269.jpeg)

As you can see the only loss that improves is the class loss, which is useless here because I only use a single class.

Unfortunately the entire dataset is relatively small, so the validation and test set are very small (around 200 images after augmentation).

Project: [Sign in to Roboflow](https://app.roboflow.com/htr1/htr1-6krjk/6)

---

<div class="post-metadata">

**Author:** ![system](https://canada1.discourse-cdn.com/flex029/uploads/roboflow1/original/1X/bb6edbfa56d0cf2ee603f061e9e09103502cc8cc.png) [@system](https://discuss.roboflow.com/u/system)\
**Post date:** [April 9, 2024, 8:30pm UTC](https://discuss.roboflow.com/t/training-loss-getting-worse-over-time-but-training-is-not-stopped/4925/2 "2024-04-09T20:30:49Z")

</div>

Closed on April 9, 2024, 8:30pm UTC

This topic was automatically closed 21 days after the last reply. New replies are no longer allowed.

---

<div class="post-metadata">

**Author:** ![Grant\_Nelson](https://yyz1.discourse-cdn.com/flex029/user_avatar/discuss.roboflow.com/grant_nelson/32/8317_2.png) [@Grant\_Nelson](https://discuss.roboflow.com/u/Grant_Nelson)\
**Post date:** [January 25, 2026, 12:54am UTC](https://discuss.roboflow.com/t/training-loss-getting-worse-over-time-but-training-is-not-stopped/4925/3 "2026-01-25T00:54:11Z")

</div>

Old thread, but for clarity for future readers, this is NOT how Roboflow chooses the final model.