---
title: README
emoji: 🔥
colorFrom: red
colorTo: indigo
sdk: static
pinned: false
---

<img src="https://raw.githubusercontent.com/asahi417/relbert/test/assets/relbert_logo.png" alt="" width="150" style="margin-left:'auto' margin-right:'auto' display:'block'"/>

<br>

RelBERT is a high-quality semantic representative embedding of word pairs powered by pre-trained language model.
Install <a href="https://pypi.org/project/relbert/">relbert</a> via pip,

<pre class="line-numbers">
   <code class="language-python">
      pip install relbert
   </code>
</pre>

and play with RelBERT models.

<pre class="line-numbers">
   <code class="language-python">
      from relbert import RelBERT
      model = RelBERT('relbert/relbert-roberta-large')
      vector = model.get_embedding(['Tokyo', 'Japan'])  # shape of (1024, )
   </code>
</pre>

See more information bellow.
<ul>
   <li>&nbsp; - GitHub: <a href="https://github.com/asahi417/relbert">https://github.com/asahi417/relbert</a></li>
   <li>&nbsp; - Paper (EMNLP 2021 main conference): <a href="https://arxiv.org/abs/2110.15705">https://arxiv.org/abs/2110.15705</a></li>
   <li>&nbsp; - HuggingFace: <a href="https://huggingface.co/relbert">https://huggingface.co/relbert</a></li>
   <li>&nbsp; - PyPI: <a href="https://pypi.org/project/relbert">https://pypi.org/project/relbert</a></li>
</ul>