---
license: mit
datasets:
- LennardZuendorf/legalis
language:
- de
metrics:
- accuracy
library_name: transformers
tags:
- legal
---

<h1> Legalis BERT Model </h1>

<h2> Model Details </h2>

<h3> Model Description </h3>

This is a court case prediction model build for a university course, this particular one utalises text classification with 

- **Developed by:** Lennard Zündorf
- **Model type:** transformer-based
- **Language(s) (NLP):** German 
- **Finetuned from model :** [German BERT/ gbert-base](https://huggingface.co/deepset/gbert-base)

<h3> Model Sources </h3>

- **Repository:** [GitHub](https://github.com/LennardZuendorf/legalis)
- **Demo:** on [Huggingface](https://huggingface.co/spaces/LennardZuendorf/legalis)

<h2> Uses </h2>

You can use this model to try and predict the outcome of a court case based on the legal facts.

<h2> Training Details </h2>

<h3>Training Data

This model uses the similarly named [dataset](https://huggingface.co/models?dataset=dataset:LennardZuendorf/legalis)

<h3> Testing Data & Metrics

<h4> Metrics </h4>

There has not been any testing yet.

<h3> Results </h3>

The accuracy score against the testing split is as high as 0.60