MSc thesis project: classification of italian certified electronic mails using SOTA machine learning, fine-tuning pre-trained deep learning models and data augmentation techniques

Official implementation of "Using Pre-Trained Language Models in an End-to-End Pipeline for Antithesis Detection" accepted in LREC-2024

Official implementation of "Using Pre-Trained Language Models in an End-to-End Pipeline for Antithesis Detection" accepted in LREC-2024

Turkish BERT/DistilBERT, ELECTRA and ConvBERT models

Industrial datasets - datasets for evaluating industrial intrusion detection systems on IPAL.

Pretrained ELECTRA Model for Korean

ML and Natual Language Processing

Simple NER Pipeline Using KoCharELECTRA

Pretrain and finetune ELECTRA with fastai and huggingface. (Results of the paper replicated !)

上海交通大学自然语言理解2020春：CoLA Task

Baseline code for Korean open domain question answering(ODQA)

Pre-training Language Models for Japanese

Solving Math Word Problems Using Language Models and Contrastive Loss

Transformers Pipeline with KoELECTRA

Albert for Conversational Question Answering Challenge

Character-level Korean ELECTRA Model (음절 단위 한국어 ELECTRA)

Sarcasm is a term that refers to the use of words to mock, irritate, or amuse someone. It is commonly used on social media. The metaphorical and creative nature of sarcasm presents a significant difficulty for sentiment analysis systems based on affective computing. The technique and results of our team, UTNLP, in the SemEval-2022 shared task 6 …

NLP 领域常见任务的实现，包括新词发现、以及基于pytorch的词向量、中文文本分类、实体识别、摘要文本生成、句子相似度判断、三元组抽取、预训练模型等。

Pre-trained Chinese ELECTRA（中文ELECTRA预训练模型）

AiSpace: Better practices for deep learning model development and deployment For Tensorflow 2.0