Hi, I am Abhijnan Nath.
AI for NLP Ph.D. Student.
Hello! I'm a Ph.D. student in CS at Colorado State University, Fort Collins, where I am advised by Nikhil Krishnaswamy . I work in AI for natural language understanding (NLU) and computational semantics, spanning topics like coreference resolution and cognate detection, Transformer-based language models, and their downstream applications with a drive to extend such research to low-resource languages. In the past, I have collaborated with the Jackson Lab, Stanford University, and the Natural Resource Ecology Lab, CSU to build their atmospheric N20 prediction pipeline using time-series data exploration.
Projects and Papers
Easy First Coreference Resolution
Don't Eat That Frog, Start with a Bagel! We sieve out lexically easier coreferents first and "learn" how to distinguish easier coreferents from the harder ones using the Longformer! In-progress.
- PyTorch
- HuggingFace
- Python
- Spacy
Loanword Detection
We detect loanwords in multiple languages including Low-resource Languages (LRL) using phonetic, semantic and articulatory features. Our work can also be extended to a language of your choice! Our paper was accepted at COLING, 2022 in Gyeoungju, South Korea.
- PyTorch
- HuggingFace
- Python
Affine Semantic Transfer
For my MS CS thesis, I explore semantic transfer using affine-mapping between large language models for two challening NLP tasks: 1) Cross Document Coreference resolution and 2) Cognate Detection. Part of this work was funded by DARPA-AIDA and orally presented at Vardial, 2022 at COLING in Gyeoungju, South Korea.
- PyTorch
- HuggingFace
- Spacy
Skills
- PyTorch
- HuggingFace
- TensorFlow
- Python
- Matlab
- AWS
- Docker