README.md

---
language: 
- en
- zh
- de
- fr
library_name: sentence-transformers
license: apache-2.0
---

# ZeroNLG

Without any labeled downstream pairs for training, ZeroNLG is an unified framework that deals with multiple natural language generation (NLG) tasks in a zero-shot manner, including image-to-text, video-to-text, and text-to-text generation tasks across English, Chinese, German, and French.

Pre-trained data: a machine-translated version of [CC3M](https://huggingface.co/datasets/conceptual_captions), including
- 1.1M English sentences 
- 1.1M English-Chinese pairs
- 1.1M English-German pairs
- 1.1M English-French pairs


Paper: [ZeroNLG: Aligning and Autoencoding Domains for Zero-Shot Multimodal and Multilingual Natural Language Generation](https://arxiv.org/abs/2303.06458)

Authors: *Bang Yang\*, Fenglin Liu\*, Yuexian Zou, Xian Wu, Yaowei Wang, David A. Clifton* 



## Quick Start
Please follow our [github repo](https://github.com/yangbang18/ZeroNLG) to prepare the environment at first.

```python
from zeronlg import ZeroNLG

# Automatically download the model from Huggingface Hub
# Note: this model is especially pre-trained for machine translation
model = ZeroNLG('zeronlg-4langs-mt')

# Translating English into Chinese
# Note: the multilingual encoder is langauge-agnostic, so the `lang` below means the langauge to be generated
output = model.forward_translate(texts='a girl and a boy are playing', lang='zh', num_beams=3)
# output = "一 个 女 孩 和 一 个 男 孩 一 起 玩"
```

## Zero-Shot Performance
### Machine translation
Model: [zeronlg-4langs-mt](https://huggingface.co/yangbang18/zeronlg-4langs-mt) only.

| En->Zh | En<-Zh | En->De | En<-De | En->Fr | En<-Fr | Zh->De | Zh<-De | Zh->Fr | Zh<-Fr | De->Fr | De<-Fr|
| :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: |
6.0|9.2|21.6|23.2|27.2|26.8|7.8|4.6|6.1|9.7|20.9|19.6


## Citation
```bibtex
@article{Yang2023ZeroNLG,
   title={ZeroNLG: Aligning and Autoencoding Domains for Zero-Shot Multimodal and Multilingual Natural Language Generation},
   author={Yang, Bang and Liu, Fenglin and Zou, Yuexian and Wu, Xian and Wang, Yaowei and Clifton, David A.},
   journal={arXiv preprint arXiv:2303.06458}
   year={2023}
}
```