# Running baselines

**URL:** <https://community.drivendata.org/t/running-baselines/4731>\
**Category:** Hateful Memes\
**Created:** [July 6, 2020, 2:15pm UTC](https://community.drivendata.org/t/running-baselines/4731 "2020-07-06T14:15:37Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![Titech008](https://avatars.discourse-cdn.com/v4/letter/t/35a633/32.png) [@Titech008](https://community.drivendata.org/u/Titech008)\
**Post date:** [July 6, 2020, 2:15pm UTC](https://community.drivendata.org/t/running-baselines/4731/1 "2020-07-06T14:15:37Z")

</div>

I am having issues running the baseline commands with mmf\_run, the command is not able to find the model/keys despite being given the location in the command. Also, giving dataset=hateful\_memes triggers a request to download the data, but when I download and unzip it in the way mentioned on the github repo then give dataset=location/of/data it does not recognize it. Is there something I should know about running this command?

---

<div class="post-metadata">

**Author:** ![slawekbiel](https://avatars.discourse-cdn.com/v4/letter/s/f19dbf/32.png) [@slawekbiel](https://community.drivendata.org/u/slawekbiel)\
**Post date:** [July 7, 2020, 7:21am UTC](https://community.drivendata.org/t/running-baselines/4731/2 "2020-07-07T07:21:40Z")

</div>

Have you run `mmf_convert_hm` as described [here](https://github.com/facebookresearch/mmf/tree/master/projects/hateful_memes#prerequisites)?

---

<div class="post-metadata">

**Author:** ![Kamyn](https://avatars.discourse-cdn.com/v4/letter/k/b5ac83/32.png) [@Kamyn](https://community.drivendata.org/u/Kamyn)\
**Post date:** [July 7, 2020, 9:13am UTC](https://community.drivendata.org/t/running-baselines/4731/3 "2020-07-07T09:13:59Z")

</div>

Try setting up this env variable MMF\_DATA\_DIR with the folder directory where your processed dataset is in.  
Something like this:  
!MMF\_DATA\_DIR=’/content/…’ mmf\_run …
