NLP

A benchmark and evaluation framework comparing instruction- vs. example-driven VLM approaches for automated content moderation policy operationalization.

Large-scale web-scraped text corpora used to train general-purpose AI models often contain harmful demographic-targeted social biases, …

We conduct a comprehensive benchmark study evaluating LLMs for demographic-targeted social bias detection in raw text data, revealing that while certain configurations show promise for scale, significant performance gaps persist across complex social categories.

Mini-project that looks at potential bias in pre-trained word embeddings and methods on how to remove such bias.

A mini-project that looks at the task of neural machine translation using sequential models and attention mechanism.

This project explores personalization of generating image captions. We explore different architecture choices of Attend2u and also analyze personalization of the captions.

This project explores a novel embedding technique for words into matrices instead of vectors. We explore this novel embedding method and how it could improve contextual sense.