# Format neuralcoref inferences for use with prodigy relations recipe

**URL:** <https://support.prodi.gy/t/format-neuralcoref-inferences-for-use-with-prodigy-relations-recipe/6041>\
**Category:** Uncategorized\
**Tags:** solved, relations, coref\
**Created:** [October 24, 2022, 3:51pm UTC](https://support.prodi.gy/t/format-neuralcoref-inferences-for-use-with-prodigy-relations-recipe/6041 "2022-10-24T15:51:17Z")\
**Posts on this page:** 1\
**Showing post:** 2

<div class="post-metadata">

**Author:** ![ryanwesslen](https://sea2.discourse-cdn.com/flex020/user_avatar/support.prodi.gy/ryanwesslen/32/2969_2.png) [@ryanwesslen](https://support.prodi.gy/u/ryanwesslen)\
**Post date:** [October 24, 2022, 4:52pm UTC](https://support.prodi.gy/t/format-neuralcoref-inferences-for-use-with-prodigy-relations-recipe/6041/2 "2022-10-24T16:52:42Z")

</div>

hi @BenGriffithsPEP!

Thanks for your message and welcome to the Prodigy community 👋

Unfortunately, we don't have an off-the-shelf "correct" recipe for `neuralcoref`. As the ticket below mentions, what would be important is to understand the format required (which I think is part of your question):

> [@How to convert the rel.manual jsonl annoatated output or export relations annotations for training the neuralcoref model](https://support.prodi.gy/t/how-to-convert-the-rel-manual-jsonl-annoatated-output-or-export-relations-annotations-for-training-the-neuralcoref-model/5720/2):
>
> I've never worked with the neuralcoref library, partially because it's a 3rd party project maintained by Huggingface. You'd need to check their documentation to understand what format they required. Have you seen the [project documentation](https://github.com/huggingface/neuralcoref/blob/master/neuralcoref/train/training.md) and [this blogpost](https://medium.com/huggingface/how-to-train-a-neural-coreference-model-neuralcoref-2-7bb30c1abdfe)?

Once you understand their format, you could create a [custom recipe](https://prodi.gy/docs/custom-recipes#writing). And as you found, it may be helpful to look over the [docs for the relations recipes](https://prodi.gy/docs/dependencies-relations#custom).

Also, you may want to look at the [`dep.correct` recipe](https://prodi.gy/docs/recipes#dep-correct) to see how a "correct" recipe with the [`"relations"` user interface](https://prodi.gy/docs/api-interfaces#relations) (note that link shows what is the format for the `"relations"` UI which may be helpful too). To view this recipe, run:

```bash
python -m prodigy stats

```

Where you should then see the `Location:` where Prodigy has been installed. In that location folder, look for the file `recipes/dep.py` where you'll see the `dep.correct` recipe.

Similarly, you may find the [`coref.manual`](https://prodi.gy/docs/recipes#coref-manual) recipe to be helpful too, which you can find in the `recipes/coref.py`. Essentially, you'd want to combine the idea of "correcting" the `neuralcoref` model into the `coref.manual` recipe.

If you're able to get a recipe, feel free to post back and/or post it as a GitHub gist! We would greatly appreciate it. Hope this helps!

---

_[View the full topic](https://support.prodi.gy/t/format-neuralcoref-inferences-for-use-with-prodigy-relations-recipe/6041)._