# de\_core\_news\_sm label question

**URL:** <https://support.prodi.gy/t/de-core-news-sm-label-question/1381>\
**Category:** Uncategorized\
**Tags:** usage\
**Created:** [April 10, 2019, 3:13pm UTC](https://support.prodi.gy/t/de-core-news-sm-label-question/1381 "2019-04-10T15:13:54Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![usteiner](https://avatars.discourse-cdn.com/v4/letter/u/edb3f5/32.png) [@usteiner](https://support.prodi.gy/u/usteiner)\
**Post date:** [April 10, 2019, 3:13pm UTC](https://support.prodi.gy/t/de-core-news-sm-label-question/1381/1 "2019-04-10T15:13:54Z")

</div>

Hi  
we want to use de\_core\_news\_sm

the following command works  
python -m prodigy ner.manual text\_prodigy\_KISME\_09042019 de\_core\_news\_sm data/text\_prodigy\_KISME\_09042019.jsonl --exclude text\_prodigy\_KISME\_09042019 --label ORG

the following command does not work  
python -m prodigy ner.manual text\_prodigy\_KISME\_09042019 de\_core\_news\_sm data/text\_prodigy\_KISME\_09042019.jsonl --exclude text\_prodigy\_KISME\_09042019 --label ORG, COMPANY

but we need two labels for annotation - which labels are allowed?

prodigy version 1.6.1  
spaCy version v2.0.16

thanks  
Uwe

---

<div class="post-metadata">

**Author:** ![ines](https://sea2.discourse-cdn.com/flex020/user_avatar/support.prodi.gy/ines/32/3_2.png) [@ines](https://support.prodi.gy/u/ines)\
**Post date:** [April 10, 2019, 3:30pm UTC](https://support.prodi.gy/t/de-core-news-sm-label-question/1381/2 "2019-04-10T15:30:14Z")

</div>

I first thought you were asking about the different annotation schemes – but I think the solution might be much simpler. Try replacing this:

> [@usteiner](#):
>
> ```python
> --label ORG, COMPANY
> 
> ```

with this:

```python
--label ORG,COMPANY

```

Spaces in command line commands usually separate arguments and values – if the values contain spaces, there's no way it can know where the value of the `label` argument ends. You can also put the labels in quotation marks, like `"ORG, COMPANY"`.

Btw, just for completeness: If you're running the `ner.manual` recipe, you'll be labelling by hand anyways and the model is only used for tokenization. So the labels that are already in the model won't matter at this step. However, if your plan is to update an existing pre-trained model, you probably want to be using consistent labels. You can find more details on the label schemes used in spaCy's pre-trained models [here](https://spacy.io/api/annotation#named-entities).

---

<div class="post-metadata">

**Author:** ![usteiner](https://avatars.discourse-cdn.com/v4/letter/u/edb3f5/32.png) [@usteiner](https://support.prodi.gy/u/usteiner)\
**Post date:** [April 10, 2019, 3:43pm UTC](https://support.prodi.gy/t/de-core-news-sm-label-question/1381/3 "2019-04-10T15:43:26Z")

</div>

thanks a lot

prodigy ner.manual text\_prodigy\_KISME\_09042019 de\_core\_news\_sm data/text\_prodigy\_KISME\_09042019.jsonl --exclude text\_prodigy\_KISME\_09042019 --label ORG,COMPANY

works fine

and yes ner.manual we use because we are at the start - as soon as we have enough data to train a modell we will shift to the other method

Kind regards  
Uwe

---

<div class="post-metadata">

**Author:** ![usteiner](https://avatars.discourse-cdn.com/v4/letter/u/edb3f5/32.png) [@usteiner](https://support.prodi.gy/u/usteiner)\
**Post date:** [April 10, 2019, 3:44pm UTC](https://support.prodi.gy/t/de-core-news-sm-label-question/1381/4 "2019-04-10T15:44:25Z")

</div>

and by the way - you are very fast with your answer - perfect service - thanks again