# Rel training **URL:** \ **Category:** Uncategorized\ **Tags:** usage, relations, training\ **Created:** [February 18, 2022, 10:14am UTC](https://support.prodi.gy/t/rel-training/5331 "2022-02-18T10:14:07Z")\ **Posts on this page:** 1\ **Showing post:** 8
**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:** [May 22, 2023, 1:14pm UTC](https://support.prodi.gy/t/rel-training/5331/8 "2023-05-22T13:14:03Z")
> [@doppio](#): > > In this situation, would you actually recommend having separate config files to build the NER and relation extractor models separately (and perhaps even a third config file to combine them into a "final model", if the relation\_extractor should only be trained on gold examples)? In general yes, we'd tend to recommend training separately. If not, you may find poor performance in `ner` may leak into the `relations` training without knowing. But if you are interested in trying to train a joint `ner-relations`, we had a post where we outlined steps to do this: > [@Training a relation extraction component](https://support.prodi.gy/t/training-a-relation-extraction-component/6376/59): > > hi @stella, Can we take a few steps back? The goal is to reproduce Sofie's rel\_component with a joint training of both ner and relations, correct? If so, I was able to reproduce without any issues. Here's my steps: Setup project and virtual environment $ git clone https://github.com/explosion/projects $ cd projects # ensure python 3.9 or 3.10 as Prodigy doesn't have wheels for 3.11 yet $ python3.9 -m venv venv $ source venv/bin/activate (venv) $ which python3 ./projects/venv/bin/python3 (v… Hope this helps! --- _[View the full topic](https://support.prodi.gy/t/rel-training/5331)._