# Error in teaching model trained with manual annotation

**URL:** <https://support.prodi.gy/t/error-in-teaching-model-trained-with-manual-annotation/2178>\
**Category:** Uncategorized\
**Tags:** usage, ner, spacy\
**Created:** [November 2, 2019, 6:33pm UTC](https://support.prodi.gy/t/error-in-teaching-model-trained-with-manual-annotation/2178 "2019-11-02T18:33:50Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![dxr](https://avatars.discourse-cdn.com/v4/letter/d/7cd45c/32.png) [@dxr](https://support.prodi.gy/u/dxr)\
**Post date:** [November 2, 2019, 6:33pm UTC](https://support.prodi.gy/t/error-in-teaching-model-trained-with-manual-annotation/2178/1 "2019-11-02T18:33:50Z")

</div>

Hi, I have a model that I batch trained after doing manual annotations. The model works reasonably well. Now I am trying to improve the model by using it in the ner.teach recipe with my custom entity.

On doing so, I am receiving an error. I am able to do a print-stream with the model and see the entity being tagged. Am I doing something wrong in the process ?

```
ERROR: Can't find label 'PRODUCT' in model product_ner 
ner.teach will only show entities with one of the specified labels.
```

---

<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:** [November 4, 2019, 2:54pm UTC](https://support.prodi.gy/t/error-in-teaching-model-trained-with-manual-annotation/2178/2 "2019-11-04T14:54:01Z")

</div>

Hi! So just to make sure I understand the problem correctly: Your `product_ner` model was trained with data containing annotations for `PRODUCT` and predicts `PRODUCT`, but when you run it with `ner.teach`, you see this error? This is indeed very strange 🤔

All the recipe really does here is check the NER component's labels (which include all labels that the model can predict). If you run the following, what does it return?

```python
import spacy

nlp = spacy.load("product_ner") # path to your model here
ner = nlp.get_pipe("ner")
print(ner.labels)

```

---

<div class="post-metadata">

**Author:** ![dxr](https://avatars.discourse-cdn.com/v4/letter/d/7cd45c/32.png) [@dxr](https://support.prodi.gy/u/dxr)\
**Post date:** [November 4, 2019, 7:54pm UTC](https://support.prodi.gy/t/error-in-teaching-model-trained-with-manual-annotation/2178/3 "2019-11-04T19:54:38Z")

</div>

Hi @ines I was able to figure out the issue, I had used one of the labels with - (hyphen) in the name. I believe hyphen got split into two tags C and PRODUCT instead of using C-PRODUCT as one label. Thanks for helping out.