Code for loralib, an implementation of "LoRA: Low-Rank Adaptation of Large Language Models"

Unattended Lightweight Text Classifiers with LLM Embeddings

Build a Large Language Model (From Scratch) book and Finetuned Models

🥇Samsung AI Challenge 2021 1등 솔루션입니다🥇

[ACL 2023] Solving Math Word Problems via Cooperative Reasoning induced Language Models (LLMs + MCTS + Self-Improvement)

DictABSA: A dictionary knowledge (entity description informations) enhanced aspect-based sentiment analysis (ABSA) code implementation

Our EMNLP 2022 paper on MCQA

금융 도메인에 특화된 한국어 임베딩 모델

The implementation of DeBERTaV3-based commonsense question answering on CommonsenseQA.

Kaggle NLP competition - Top 2% solution (36/2060)

Application for training the pretrained transformer model DeBERTaV3 on an Aspect Based Sentiment Analysis task

A collection of my Kaggle competition solutions.

Key Point Analysis: implementation of two-component system for performing Key Point Matching and Key Point Generation task with multiple PLMs.

The code of Hierarchical Multi-label Classification (HMC). It is a final course project of Natural Language Processing and Deep Learning, 2022 Fall.

Deberta implementation at Cross-Domain Sentiment Classification With Bidirectional Contextualized Transformer Language Models

Implementing science-related multiple-choice question answering based on LLMs and RAG.

Scraping paper data, preprocessed and trained using BERT variants, deployment and an integration to website

Finetuning Large Language Models

Classification of medical texts to differentiate between human medical and veterinary subjects.

71M parameter safety classifier (DeBERTa-v3-xsmall). Dual-head: binary safe/unsafe + 7 category labels. Sub-2ms inference.

Code release for KCLarity at SemEval-2026 Task 6: Encoder and Zero-Shot Approaches to Political Evasion Detection

Real-time LLM hallucination guardrail — NLI + RAG fact-checking with token-level streaming halt. Drop-in for any LLM backend.

This repository contains the code for submission made at SemEval 2022 Task 5: MAMI

Can you spot automatically generated scientific excerpts?

Backbone: 5 x DeBerta and Head: Rapids SVR

This repository presents a complete deep learning pipeline for detecting extremist content in social media text.

141M param safety model (not much better than v1, but a great learning)

Advanced financial NLP system developed within the Cambridge × Bank of England program. Extracts supervisory risk signals from JP Morgan & HSBC earnings call Q&A using LLMs, BERTopic and sentiment modelling.

Built a real-time multimodal AI system to detect and analyze disaster-related misinformation on social media. The project fuses textual, visual, and network propagation signals to provide accurate and timely detection of false information during critical events.

Authorship attribution model built for an internal ML competition at Endava. I fine-tuned a DeBERTa transformer and achieved 0.24 log loss, ranking ~30 out of 300 participants. Focused on model generalization, data cleaning, and avoiding known data leakage techniques.