# What is the training runtime?

**URL:** <https://community.drivendata.org/t/what-is-the-training-runtime/11021>\
**Category:** Goodnight Moon, Hello Early Literacy Screening\
**Created:** [December 3, 2024, 5:06am UTC](https://community.drivendata.org/t/what-is-the-training-runtime/11021 "2024-12-03T05:06:46Z")\
**Posts on this page:** 6\
**Page:** 1

<div class="post-metadata">

**Author:** ![dylanliu](https://avatars.discourse-cdn.com/v4/letter/d/ac91a4/32.png) [@dylanliu](https://community.drivendata.org/u/dylanliu)\
**Post date:** [December 3, 2024, 5:06am UTC](https://community.drivendata.org/t/what-is-the-training-runtime/11021/1 "2024-12-03T05:06:46Z")

</div>

Since the winners solutions will be trained again, I want to know, is the training runtime the same as the test runtime?

---

<div class="post-metadata">

**Author:** ![emily](https://avatars.discourse-cdn.com/v4/letter/e/8c91f0/32.png) [@emily](https://community.drivendata.org/u/emily)\
**Post date:** [December 3, 2024, 5:29pm UTC](https://community.drivendata.org/t/what-is-the-training-runtime/11021/2 "2024-12-03T17:29:21Z")

</div>

There is not a specified training runtime. Prize finalists will be given more detail on packaging up their code at the end of the competition. For now, the biggest thing to pay attention to is keeping training code reproducible (this includes tracking any additional package dependencies as well).

---

<div class="post-metadata">

**Author:** ![Kennykenster](https://yyz2.discourse-cdn.com/flex028/user_avatar/community.drivendata.org/kennykenster/32/1824_2.png) [@Kennykenster](https://community.drivendata.org/u/Kennykenster)\
**Post date:** [December 6, 2024, 7:40pm UTC](https://community.drivendata.org/t/what-is-the-training-runtime/11021/3 "2024-12-06T19:40:47Z")

</div>

So this means there is no limit on the (re-)training runtime of the final model? Is there any possibility that this will be changed later on, considering this would have a major impact on code reproducibility?

---

<div class="post-metadata">

**Author:** ![emily](https://avatars.discourse-cdn.com/v4/letter/e/8c91f0/32.png) [@emily](https://community.drivendata.org/u/emily)\
**Post date:** [December 6, 2024, 8:14pm UTC](https://community.drivendata.org/t/what-is-the-training-runtime/11021/4 "2024-12-06T20:14:03Z")

</div>

@Kennykenster can you share more specifically what your concern is? For example, are you worried about time limits or specific compute constraints?

---

<div class="post-metadata">

**Author:** ![Kennykenster](https://yyz2.discourse-cdn.com/flex028/user_avatar/community.drivendata.org/kennykenster/32/1824_2.png) [@Kennykenster](https://community.drivendata.org/u/Kennykenster)\
**Post date:** [December 8, 2024, 4:19pm UTC](https://community.drivendata.org/t/what-is-the-training-runtime/11021/5 "2024-12-08T16:19:08Z")

</div>

Yes, are any specific details on training-time limits and are there further hardware specifications?

---

<div class="post-metadata">

**Author:** ![emily](https://avatars.discourse-cdn.com/v4/letter/e/8c91f0/32.png) [@emily](https://community.drivendata.org/u/emily)\
**Post date:** [December 9, 2024, 3:05am UTC](https://community.drivendata.org/t/what-is-the-training-runtime/11021/6 "2024-12-09T03:05:58Z")

</div>

@Kennykenster there aren’t any specific details on training-time limits or further hardware specifications at this time.