README.md

---
language:
- fa
library_name: hezar
tags:
- feature-extraction
- hezar
pipeline_tag: feature-extraction
---
This is the original fasttext embedding model for Persian from [here](https://fasttext.cc/docs/en/crawl-vectors.html#models) loaded and converted using Gensim and exported to Hezar compatible format.
For more info, see [here](https://fasttext.cc/docs/en/support.html).

In order to use this model in Hezar you can simply use this piece of code:
```bash
pip install hezar
```
```python
from hezar.embeddings import Embedding

fasttext = Embedding.load("hezarai/fasttext-fa-300")
# Get embedding vector
vector = fasttext("هزار")
# Find the word that doesn't match with the rest
doesnt_match = fasttext.doesnt_match(["خانه", "اتاق", "ماشین"])
# Find the top-n most similar words to the given word
most_similar = fasttext.most_similar("هزار", top_n=5)
# Find the cosine similarity value between two words
similarity = fasttext.similarity("مهندس", "دکتر")
```