# Issues with ner.batch-train with en\_trf\_bertbaseuncased\_lg after creating a custom set of labels

**URL:** <https://support.prodi.gy/t/issues-with-ner-batch-train-with-en-trf-bertbaseuncased-lg-after-creating-a-custom-set-of-labels/2101>\
**Category:** Uncategorized\
**Tags:** enhancement, usage, ner, solved, transformers\
**Created:** [October 13, 2019, 9:51pm UTC](https://support.prodi.gy/t/issues-with-ner-batch-train-with-en-trf-bertbaseuncased-lg-after-creating-a-custom-set-of-labels/2101 "2019-10-13T21:51:37Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![darrengarvey](https://avatars.discourse-cdn.com/v4/letter/d/848f3c/32.png) [@darrengarvey](https://support.prodi.gy/u/darrengarvey)\
**Post date:** [October 13, 2019, 9:51pm UTC](https://support.prodi.gy/t/issues-with-ner-batch-train-with-en-trf-bertbaseuncased-lg-after-creating-a-custom-set-of-labels/2101/1 "2019-10-13T21:51:37Z")

</div>

Hey all.

I started by creating a dataset using `prodigy dataset ...` and then using `prodigy ner.manual ...` with a bunch of my own labels I annotated a bunch of examples.

I was initially planning on using the BERT model: en\_trf\_bertbaseuncased\_lg, but trying to run batch-train with:

`prodigy ner.batch-train demo_v01 en_trf_bertbaseuncased_lg --output /tmp/model --eval-split 0.2 --dropout 0.2`

I got the following error:

`KeyError: "[E001] No component 'trf_tok2vec' found in pipeline. Available names: ['sentencizer', 'ner']"`

Is there some missing import in prodigy?

---

<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:** [October 14, 2019, 8:58am UTC](https://support.prodi.gy/t/issues-with-ner-batch-train-with-en-trf-bertbaseuncased-lg-after-creating-a-custom-set-of-labels/2101/2 "2019-10-14T08:58:47Z")

</div>

Hi! We currently do not have an NER model implementation using the transformer weights. See here for details:

> <https://github.com/explosion/spacy-transformers/issues/23>
>
> Hi,
> I am wondering whether the current codebased support ner pipeline from spacy, thanks.

So running a transformer model with `ner.batch-train` doesn't really make sense – you'd always be training a regular spaCy NER model (so you might as well use a blank `en` model). To use the transformer models with Prodigy likely also require slightly modified training recipes, since the updating works slightly differently in those cases (and has additional configuration options).