# How to train a model on multiple datasets

**URL:** <https://discuss.huggingface.co/t/how-to-train-a-model-on-multiple-datasets/55469>\
**Category:** Beginners\
**Created:** [September 18, 2023, 2:25pm UTC](https://discuss.huggingface.co/t/how-to-train-a-model-on-multiple-datasets/55469 "2023-09-18T14:25:40Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![Altabus](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/altabus/32/21087_2.png) [@Altabus](https://discuss.huggingface.co/u/Altabus)\
**Post date:** [September 18, 2023, 2:25pm UTC](https://discuss.huggingface.co/t/how-to-train-a-model-on-multiple-datasets/55469/1 "2023-09-18T14:25:40Z")

</div>

I was able to make a translation model as shown in the manual, but the quality is not very good, I wanted to feed it several datasets, but I realized that they do not add to each other, but overwrite the model, so how can I feed it several datasets?

And in general, did I decide to solve this problem correctly?

---

<div class="post-metadata">

**Author:** ![dblakely](https://avatars.discourse-cdn.com/v4/letter/d/54ee81/32.png) [@dblakely](https://discuss.huggingface.co/u/dblakely)\
**Post date:** [September 18, 2023, 3:41pm UTC](https://discuss.huggingface.co/t/how-to-train-a-model-on-multiple-datasets/55469/2 "2023-09-18T15:41:00Z")

</div>

Huggingface has an [interleave datasets function](https://huggingface.co/docs/datasets/v2.14.5/en/package_reference/main_classes#datasets.interleave_datasets) you could check out to combine several datasets together.

> And in general, did I decide to solve this problem correctly?

Using more data very well might help, but hard to say without more context. Lots of things can make a model good or bad.