# Train BERT on time-series data

**URL:** <https://discuss.huggingface.co/t/train-bert-on-time-series-data/3697>\
**Category:** Models\
**Created:** [February 17, 2021, 8:10pm UTC](https://discuss.huggingface.co/t/train-bert-on-time-series-data/3697 "2021-02-17T20:10:35Z")\
**Posts on this page:** 1\
**Showing post:** 2

<div class="post-metadata">

**Author:** ![aclifton314](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/aclifton314/32/55465_2.png) [@aclifton314](https://discuss.huggingface.co/u/aclifton314)\
**Post date:** [September 7, 2022, 5:20pm UTC](https://discuss.huggingface.co/t/train-bert-on-time-series-data/3697/2 "2022-09-07T17:20:36Z")

</div>

@clems I am running a similar experiment currently and have posted my thoughts on it [here](https://discuss.huggingface.co/t/using-time-series-for-sequenceclassification-models/3289). The largest difference I have to yours is that I don’t have to run the self-supervised training that BERT did, as I have labels. Since it seems you are running the self-supervised training, were you able to obtain any results with your initial suggestion? I’d be interested to hear your findings.

---

_[View the full topic](https://discuss.huggingface.co/t/train-bert-on-time-series-data/3697)._