vijay viswanathan

(he/him)

PhD student, software engineer, scientist in training

Hi!

I'm a PhD student at the Language Technologies Institute in CMU's School of Computer Science, advised by Sherry Tongshuang Wu and Graham Neubig. I'm interested in developing interfaces for anyone to build and correct NLP models. Previously I was generously supported by an NEC Student Research Fellowship.

In 2022, I received my MS in Intelligent Information Systems from CMU LTI, working with Pengfei Liu and Graham.

Before grad school, I worked as a software engineer for 4 years, deploying machine learning to solve practical problems at small startups (Yik Yak, Scaled Inference) and at a large industrial R&D lab (Uber ATG).

In 2016, I received a bachelor's degree in mathematics from CMU's Mellon College of Science.

News

(03-20-22) I'll be staying at CMU for the next 3+ years as a PhD student in Language and Information Technologies.
(06-07-21) Started a summer internship with Yoav Goldberg at AI2 Israel (remote) working on biomedical relation extraction!
(05-06-21) First-ever conference paper accepted (at ACL-IJCNLP 2021)! We show that using context from the citation graph is valuable for scientific information extraction.
(8-25-20) Joined the the MIIS program at CMU.

Publications

Self-Guide: Better Task-Specific Instruction Following via Self-Synthetic Finetuning (code, bibtex)
Chenyang Zhao*, Xueying Jia*, Vijay Viswanathan, Tongshuang Wu, Graham Neubig
COLM 2024
Training Task Experts through Retrieval Based Distillation (code, bibtex)
Jiaxin Ge*, Xueying Jia*, Vijay Viswanathan, Hongyin Luo, Graham Neubig
arXiv (2024)
Synthetic Multimodal Question Generation (bibtex)
Ian Wu*, Sravan Jayanthi *, Vijay Viswanathan, Simon Rosenberg, Sina Pakazad, Tongshuang Wu, Graham Neubig
arXiv (2024)
Better Synthetic Data by Retrieving and Transforming Existing Datasets (code, bibtex)
Saumya Gandhi*, Ritu Gala*, Vijay Viswanathan, Tongshuang Wu, Graham Neubig
ACL Findings 2024
Prompt2Model: Generating Deployable Models from Natural Language Instructions (video, code, bibtex)
Vijay Viswanathan*, Chenyang Zhao*, Amanda Bertsch, Tongshuang Wu, Graham Neubig
EMNLP 2023 Demo Track
Large Language Models Enable Few-Shot Clustering (code, bibtex)
Vijay Viswanathan, Kiril Gashteovski, Carolin Lawrence, Tongshuang Wu, Graham Neubig
Transactions of the Association for Computational Linguistics (2024)
DataFinder: Scientific Dataset Recommendation from Natural Language Descriptions (code, bibtex)
Vijay Viswanathan, Luyu Gao, Tongshuang Wu, Pengfei Liu, Graham Neubig
ACL 2023
A Dataset for N-ary Relation Extraction of Drug Combinations (video, code, bibtex)
Aryeh Tiktinsky*, Vijay Viswanathan*, Danna Niezni, Dana Meron Azagury, Yosi Shamay, Hillel Taub-Tabib, Tom Hope, Yoav Goldberg
NAACL 2022 (oral)
DataLab: A Platform for Data Analysis and Intervention (demo, code, bibtex)
Yang Xiao, Jinlan Fu, Weizhe Yuan, Vijay Viswanathan, Zhoumianze Liu, Yixin Liu, Graham Neubig, Pengfei Liu
ACL 2022 Demo Track. [Outstanding Demo Paper]
Rethinking End-to-End Evaluation of Decomposable Tasks: A Case Study on Spoken Language Understanding (code, bibtex)
Siddhant Arora*, Alissa Ostapenko*, Vijay Viswanathan*, Siddharth Dalmia*, Florian Metze, Shinji Watanabe, Alan W Black
Interspeech 2021
CitationIE: Leveraging the Citation Graph for Scientific Information Extraction (video, code, bibtex)
Vijay Viswanathan, Graham Neubig, Pengfei Liu
ACL-IJCNLP 2021 (oral)

Academic Service

Member of CMU LTI DEI Committee (2022-2023).
Co-organizer of the 2023 CMU LTI Seminar on Large Language Models
Teaching Assistant for 11-737: Multilingual NLP (Spring 2022)
Student Volunteer for LTI Faculty Hiring Committee (Spring 2021, Spring 2022, Spring 2023). (Chaired the student committee in Spring 2022)
Reviewer for NeurIPS CtrlGen workshop (2021)

Consulting

I occasionally get emails from founders or teams in industry looking to chat about ideas. I would love to help, but unfortunately do not have the bandwidth to participate in all such discussions. Accordingly, I'm now offering consulting at a rate of $300/hour. Feel free to book an initial consultation to see if this could be a good fit.

Expertise: synthetic data, large language models, dataset curation and annotation, clustering/exploratory analysis of documents