I'm a mathematics undergraduate at Baruch College (CUNY) interested in NLP and the evaluation of language models. My work focuses on understanding model behavior and the limits of how we measure it: where standard metrics succeed, where they fail, and what more rigorous evaluation might look like.
-
Fine-Tuning LLaMA-2-7B-Chat for Structured Cover Letter Generation
QLoRA supervised fine-tuning pipeline with quantitative and qualitative evaluation
→ https://github.com/soniatyburczy/llama2-qlora-finetune -
Structural Predictors of Chronic Absenteeism in NYC Public Schools
RF and XGBoost regression models identifying structural predictors of chronic absenteeism across NYC public schools.
→ https://github.com/soniatyburczy/absenteeism-structural-predictors
- Website: https://soniatyburczy.github.io
- Hugging Face: https://huggingface.co/czszt
- LinkedIn: https://www.linkedin.com/in/sonia-tyburczy-427309313/