R&D Activities

Organizing NLP challenges

Poleval

PolEval is a SemEval – inspired evaluation campaign for natural language processing tools for Polish. Submitted tools compete against one another within certain tasks selected by organizers, using available data and are evaluated according to pre-established procedures.

1. PolEval2021 – Task 1: Punctuation restoration from read text.

2. PolEval2022 – in progress

Publications

1. Submitted paper: Piotr Pęzik, Agnieszka Mikołajczyk, Adam Wawrzyński, Bartłomiej Nitoń, Maciej Ogrodniczuk. ‘Keyword Extraction with a Text-To-Text Transfer Transformer (T5)‘, ACIIDS, 2022.

2. Pappagari, Raghavendra, Piotr Żelasko, Agnieszka Mikołajczyk, Piotr Pęzik, and Najim Dehak. ‘Joint Prediction of Truecasing and Punctuation for Conversational Speech in Low-Resource Scenarios’, 2021. https://arxiv.org/abs/2109.06103.

3. Piotrowski, Mateusz, Wojciech Janowski, and Piotr Pęzik. ‘A Bidirectional LSTM-CRF Network with Subword Representations, Character Convolutions and Morphosyntactic Features for Named Entity Recognition in Polish’. Proceedings of the PolEval 2018 Workshop, 2018, 93.

4. Lew, Michał, and Piotr Pęzik. ‘A Sequential Child-Combination Tree-LSTM Network for Sentiment Analysis’. In Human Language Technologies as a Challenge for Computer Science and Linguistics, 397–401. Poznań, 2017

5. Pęzik, Piotr, and Sebastian Laskowski. ‘Evaluating an Averaged Perceptron Morphosyntactic Tagger for Polish’. In Human Language Technologies as a Challenge for Computer Science and Linguistics, 372–76. Poznań, 2017

6. Agnieszka Mikołajczyk, Wojciech Janowski, Adam Wawrzyński, Piotr Pęzik, Michał Adamczyk, Adam Kaczmarek. ‘Punctuation Restoration from Read Text’, 22–32, n.d.

Conferences

1. Filip Żarnecki, Michał Laskowski, Maciej Rogala. 'How do we feel about COVID-19 vaccinations?’. ML in PL Conference Presentation, November 2021

2. Agnieszka Mikołajczyk, Wojciech Janowski, Adam Wawrzyński, Piotr Pęzik, Michał Adamczyk, Adam Kaczmarek. ‘Punctuation Restoration from Read Text’, PolEval2021, October 2021

Voicelab Summer Camp

R&D projects for Students

NLP team support

Access to our infrastructure: GPU clusters (e.g. 8 x Quadro RTX 8000/NVIDIA GeForce RTX 2080 Ti)

more details soon …

is here! 🎉

VoiceLab.AI, leader in Conversational AI now brings TRURL, an instruction-following large language model (LLM) which has been fine-tuned for number of business domains such as e-commerce and customer support.

Vencode harnesses TRURL to build a company chat system, seamlessly integrating information from provided documents and the website for enhanced communication within the organization.