Installation

VizSeq requires Python 3.11+ and currently runs on Unix/Linux and macOS/OS X. It will support Windows as well in the future.

You can install VizSeq from PyPI repository:

$ pip install vizseq

Or install it from source:

$ git clone https://github.com/facebookresearch/vizseq
$ cd vizseq
$ pip install -e .

Optional Dependencies

The base install keeps the dependency footprint small. It covers all the n-gram-based metrics, both the Jupyter Notebook and Web App interfaces, and the Fairseq integration. Heavier features live behind extras:

Extra	Install	Enables
embeddings	pip install "vizseq[embeddings]"	Embedding-based metrics: LASER and BERTScore (pulls in PyTorch and Transformers)
audio	pip install "vizseq[audio]"	Audio sources for speech recognition and speech translation tasks
translate	pip install "vizseq[translate]"	Google Translate integration
all	pip install "vizseq[all]"	All of the above

Extras can be combined, and they work with source installs too:

$ pip install "vizseq[embeddings,audio]"
$ pip install -e ".[all]"

vizseq.available_scorers() lists every built-in scorer regardless of which extras are installed. Scorers that need an extra raise an ImportError naming the extra when you actually use them, so bert_score and laser are listed even without the embeddings extra.

Citation

If you find VizSeq useful in your research, please cite as

@inproceedings{wang2019vizseq,
  title = {VizSeq: A Visual Analysis Toolkit for Text Generation Tasks},
  author = {Wang, Changhan and Jain, Anirudh and Chen, Danlu and Gu, Jiatao},
  booktitle = {In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing: System Demonstrations},
  year = {2019},
}