Spaces:

AUEB-NLP

/

greek-nlp-toolkit-demo

like 2

Running on Zero

Hi there, we are the AUEB NLP Group (Athens University of Economics and Business). We are building the state-of-the-art and open-source NLP toolkit for natural language processing tasks, regarding the Greek Language.

Our software has been built upon recent research of different BSc theses and papers, like:

1. C. Dikonimaki, "A Transformer-based natural language processing toolkit for Greek -- Part of speech tagging and dependency parsing", BSc thesis, Department of Informatics, Athens University of Economics and Business, 2021. http://nlp.cs.aueb.gr/theses/dikonimaki_bsc_thesis.pdf

2. N. Smyrnioudis, "A Transformer-based natural language processing toolkit for Greek -- Named entity recognition and multi-task learning", BSc thesis, Department of Informatics, Athens University of Economics and Business, 2021. http://nlp.cs.aueb.gr/theses/smyrnioudis_bsc_thesis.pdf

3. Toumazatos, A., Pavlopoulos, J., Androutsopoulos, I., & Vassos, S. (2024). Still All Greeklish to Me: Greeklish to Greek Transliteration. In Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024) (pp. 15309–15319).

We have integrated the upon research projects into one Python toolkit called gr-nlp-toolkit, which was also partially funded by Google's Summer of Code, a 12-week program for open-source projects (https://ellak.gr/wiki/index.php?title=Google_Summer_of_Code_2024_proposed_ideas#Greeklish-to-Greek:_Development_of_an_open-source_and_state-of-the-art_toolkit).

Building upon all those recent works, we are currently in the process of creating a System Demonstration paper so people can learn more about this cool open-source toolkit. Right now, we have built a demo for the community to be able to play with it without using any code (https://huggingface.co/spaces/AUEB-NLP/greek-nlp-toolkit-demo/), as well as an API (https://huggingface.co/spaces/AUEB-NLP/The-Greek-NLP-API), all hosted in HuggingFace, so all researchers and practitioners can access it even if they do not use Python.

Having all that said, we would love to have some GPU credits for you, since inference time is a big issue. Right now, the CPU inference time is over 30 seconds, making our HuggingFace space a bit unattractive. Of course, we would love to acknowledge your GPU grant in our demonstration paper, which, by the way, will be sent to a top-tier NLP venue like COLING or EACL.

Thank you very much.