ZeroShot XNLI

Repository accompanying the CLEF 2020 eRisk Workshop Working Notes for the iLab team (University Of Strathclyde)

A multilingual model XLM- RoBERTa for the textual entailment of sequence pair - premise and hypothesis of 15 different languages using the MNLI and XNLI corpus.

NLP Workshop -ML India

Our source code for EACL2021 workshop: Offensive Language Identification in Dravidian Languages. We ranked 4th, 4th and 3rd in Tamil, Malayalam and Kannada language of this task finally!🥳

Sentiment Analysis of tweets written in underused Slavic languages (Serbian, Bosnian and Croatian) using pretrained multilingual RoBERTa based model XLM-R on 2 different datasets.

🥉 (Bronze medal - 100th place - Top 7%) Repository for the "Jigsaw Multilingual Toxic Comment Classification" Kaggle competition.

The winning system of the Text2TCS project team submitted to the CogaLex VI shared task.

This is a Pytorch (+ Huggingface transformers) implementation of a "simple" text classifier defined using BERT-based models. In this lab we will see how it is simple to use BERT for a sentence classification task, obtaining state-of-the-art results in few lines of python code.

An implementation of drophead regularization for pytorch transformers

Fine-tuning Question Answering models on German with the GermanQuAD dataset

A case study of NLI ( Natural Language Inferencing) with Transfer Learning. Kaggle Competition Rank - 18th (Global)

NLP notebooks

HASOC2021: Subtask 2 a) Codemix Challenge; Contains baselines and hierarchical approach that extracts the relevant context useful for classification of hostile tweets on English-Hindi code-mix data obtained from twitter.

Resources and tools for the Tutorial - "Hate speech detection, mitigation and beyond" presented at ICWSM 2021

A Transformer-based Sequence-labeling Approach to the Slovenian Cross-domain Automatic Term Extraction

Clickbait detector for English tweets, trained on Webis-17 dataset

Deep-learning system proposed by HFL for SemEval-2022 Task 8: Multilingual News Similarity

1st place solution to AI IJC Customer Service task

Training and evaluating topic classification models (fastText and Transformer-based language models) for topic classification of Slovenian news texts. The repository can be used as a tutorial to learn topic classification.