EmoTa is an open-access Tamil Speech Emotion Recognition dataset with 936 utterances from 22 native speakers, covering five emotions (anger, happiness, sadness, fear, and neutrality). It supports emotion classification tasks and advances Tamil language processing.

GraCoRe: Benchmarking Graph Comprehension and Complex Reasoning in Large Language Models（Published in COLING 2025）

[COLING 2025] CPsyExam: A Chinese Benchmark for Evaluating Psychology using Examinations

Revisiting Cosine Similarity via Normalized ICA-transformed Embeddings (Published in COLING 2025)

[COLING 2025] Official PyTorch Implementation of "Decoding Decoded: Understanding Hyperparameter Effects in Open-Ended Text Generation"

Official repository for the paper 'From Multiple-Choice to Extractive QA: A Case Study for English and Arabic' (COLING 2025). Includes datasets, scripts, and benchmarks for converting MCQA to EQA across English, Modern Standard Arabic, and Arabic dialects

Norm of Mean Contextualized Embeddings Determines their Variance (Published in COLING 2025)

[COLING 2025] VaeDiff-DocRE: End-to-end Data Augmentation for Document-level Relation Extraction via Variational Autoencoder and Diffusion Prior

Code for 《Ambiguity-aware Multi-level Incongruity Fusion Network for Multi-Modal Sarcasm Detection》.

Morphology-enhanced neural models for Ancient Greek interlinear translation, achieving 35-38% BLEU improvements for English and Polish translations. Includes custom T5 implementations and training code. [LoResLM@COLING2025]

This is the implementation for my paper on GenAIK workshop at COLING 2025.

Relevant code and data for the COLING2025 paper: Evaluating LLMs' Capability to Identify Lexical Semantic Equivalence: Probing with the Word-in-Context Task