README.md

---
license: cc-by-nc-4.0
---

# Model Card for Spivavtor-Large

This model was obtained by fine-tuning the corresponding `bigscience/mt0-large` model on the Spivavtor dataset. All details of the dataset and fine tuning process can be found in our paper and repository.

**Paper:** Spivavtor: An Instruction Tuned Ukrainian Text Editing Model

**Authors:** Aman Saini, Artem Chernodub, Vipul Raheja, Vivek Kulkarni

## Model Details

### Model Description

- **Language**: Ukrainian
- **Finetuned from model:** bigscience/mt0-large

## How to use
We make available the following models presented in our paper. 

<table>
  <tr>
    <th>Model</th>
    <th>Number of parameters</th>
    <th>Reference name in Paper</th>
  </tr>
  <tr>
    <td>Spivavtor-large</td>
    <td>1.2B</td>
    <td>Spivavtor-mt0-large</td>
  </tr>
  <tr>
    <td>Spivavtor-xxl</td>
    <td>11B</td>
    <td>Spivavtor-aya-101</td>
  </tr>  
</table>

## Usage
```python
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
tokenizer = AutoTokenizer.from_pretrained("grammarly/spivavtor-large")
model = AutoModelForSeq2SeqLM.from_pretrained("grammarly/spivavtor-large")

input_text = 'Виправте граматику в цьому реченнi: Дякую за iнформацiю! ми з Надiєю саме вийшли з дому'

input_ids = tokenizer(input_text, return_tensors="pt").input_ids
outputs = model.generate(input_ids, max_length=256)
output_text = tokenizer.decode(outputs[0], skip_special_tokens=True)
```