Junlin Wang

Duke University

junlin.wang2[at]duke.edu

[Google Scholar]
[Github]
[Projects]
[CV]

About Me

I am a 3rd year Computer Science PhD at Duke Unviersity, advised by Bhuwan Dhingra and previously also advised by Sam Wiseman from 2022-2023.

Before Duke, I work closely with Sameer Singh on Machine Learning Interpretaions and Natural Language Processing Projects. I also worked as a research intern in Together AI, AWS, Intel, Tencent and Applied AI lab at Comcast.

My research experience has been prmiarily on LLM reasoning, agents, and alignment.

News

I will be presenting at EMNLP 2024!

publications

Mixture-of-Agents Enhances Large Language Model Capabilities

Junlin Wang, Jue Wang, Ben Athiwaratkun, Ce Zhang, James Zou
arXiv
bib code blog

@misc{wang2024mixtureofagentsenhanceslargelanguage,
                title={Mixture-of-Agents Enhances Large Language Model Capabilities}, 
                author={Junlin Wang and Jue Wang and Ben Athiwaratkun and Ce Zhang and James Zou},
                year={2024},
                eprint={2406.04692},
                archivePrefix={arXiv},
                primaryClass={cs.CL},
                url={https://arxiv.org/abs/2406.04692}, 
              }

Reasoning in Token Economies: Budget-Aware Evaluation of LLM Reasoning Strategies

Junlin Wang, Siddhartha Jain, Dejiao Zhang, Baishakhi Ray, Varun Kumar, Ben Athiwaratkun.
EMNLP 2024
bib

@article{wang2024reasoning,
                title   = {Reasoning in Token Economies: Budget-Aware Evaluation of LLM Reasoning Strategies},
                author  = {Junlin Wang and Siddhartha Jain and Dejiao Zhang and Baishakhi Ray and Varun Kumar and Ben Athiwaratkun},
                year    = {2024},
                journal = {arXiv preprint arXiv: 2406.06461}
              }

ReCaLL: Membership Inference via Relative Conditional Log-Likelihoods

Roy Xie, Junlin Wang, Ruomin Huang, Minxing Zhang, Rong Ge, Jian Pei, Neil Zhenqiang Gong, Bhuwan Dhingra.
EMNLP 2024
bib code blog

@article{xie2024recall,
                title   = {ReCaLL: Membership Inference via Relative Conditional Log-Likelihoods},
                author  = {Roy Xie and Junlin Wang and Ruomin Huang and Minxing Zhang and Rong Ge and Jian Pei and Neil Zhenqiang Gong and Bhuwan Dhingra},
                year    = {2024},
                journal = {arXiv preprint arXiv: 2406.15968}
              }

Raccoon: Prompt Extraction Benchmark of LLM-Integrated Applications

Junlin Wang*, Tianyi Yang*, Roy Xie, Bhuwan Dhingra.
ACL 2024 Findings
bib code

@article{wang2024raccoon,
                title   = {Raccoon: Prompt Extraction Benchmark of LLM-Integrated Applications},
                author  = {Junlin Wang and Tianyi Yang and Roy Xie and Bhuwan Dhingra},
                year    = {2024},
                journal = {arXiv preprint arXiv: 2406.06737}
              }

LLM-Resistant Math Word Problem thumbnail

LLM-Resistant Math Word Problem Generation via Adversarial Attacks

Roy Xie, Chengxuan Huang, Junlin Wang, Bhuwan Dhingra.
EMNLP 2024 Findings
bib code

@inproceedings{Xie2024adversarial,
                title  = {Adversarial Math Word Problem Generation},
                author = {Roy Xie and Chengxuan Huang and Junlin Wang and Bhuwan Dhingra},
                year   = {2024},
                url    = {https://openreview.net/forum?id=bJz5uGzEe6&referrer=%5BAuthor%20Console%5D(%2Fgroup%3Fid%3Daclweb.org%2FACL%2FARR%2F2024%2FJune%2FAuthors%23your-submissions)}
              }

NeuroComparatives: Neuro-Symbolic Distillation of Comparative Knowledge

Phillip Howard*, Junlin Wang*, Vasudev Lal, Gadi Singer, Yejin Choi, Swabha Swayamdipta.
NAACL 2024 Findings
bib

@misc{howard2023neurocomparatives,
                title={NeuroComparatives: Neuro-Symbolic Distillation of Comparative Knowledge}, 
                author={Phillip Howard and Junlin Wang and Vasudev Lal and Gadi Singer and Yejin Choi and Swabha Swayamdipta},
                year={2023},
                eprint={2305.04978},
                archivePrefix={arXiv},
                primaryClass={cs.CL}
          }

Maestro: A Gamified Platform for Teaching AI Robustness

Margarita Geleta, Jiacen Xu, Manikanta Loya, Junlin Wang, Sameer Singh, Zhou Li and Sergio Gago Masague.
EAAI 2023
bib

@inproceedings{DBLP:conf/aaai/GeletaXLW00M23,
                author    = {Margarita Geleta and Jiacen Xu and Manikanta Loya and Junlin Wang and Sameer Singh and Zhou Li and Sergio Gago Masagu{\'{e}}},
                editor    = {Brian Williams and Yiling Chen and Jennifer Neville},
                title     = {Maestro: {A} Gamified Platform for Teaching {AI} Robustness},
                booktitle = {Thirty-Seventh {AAAI} Conference on Artificial Intelligence, {AAAI} 2023, Thirty-Fifth Conference on Innovative Applications of Artificial Intelligence, {IAAI} 2023, Thirteenth Symposium on Educational Advances in Artificial Intelligence, {EAAI} 2023, Washington, DC, USA, February 7-14, 2023},
                pages     = {15816-15824},
                publisher = {{AAAI} Press},
                year      = {2023},
                url       = {https://doi.org/10.1609/aaai.v37i13.26878},
                doi       = {10.1609/AAAI.V37I13.26878},
                timestamp = {Sun, 12 Nov 2023 02:11:30 +0100},
                biburl    = {https://dblp.org/rec/conf/aaai/GeletaXLW00M23.bib},
                bibsource = {dblp computer science bibliography, https://dblp.org}
              }

Gradient-based Analysis of NLP Models is Manipulable

Junlin Wang*, Jens Tuyls*, Eric Wallace and Sameer Singh.
EMNLP 2020 Findings
We found that gradient-based interpretation techniques are not faithful
bib blog

@inproceedings{wang2020gradientbased,
                      Author = {Junlin Wang, Jens Tuyls, Eric Wallace and Sameer Singh},
                      Booktitle = {Empirical Methods in Natural Language Processing Findings},                            
                      Year = {2020},
                      Title = {Gradient-based Analysis of NLP Models is Manipulable}}

AllenNLP Interpret: A Framework for Explaining Predictions of NLP Models

Eric Wallace, Jens Tuyls, Junlin Wang, Sanjay Subramanian, Matt Gardner, and Sameer Singh.
Demo at EMNLP 2019 Best Demo Award
An open-source toolkit built on top of AllenNLP that makes it easy to interpret NLP models.
bib blog

@inproceedings{Wallace2019AllenNLP,
                      Author = {Eric Wallace and Jens Tuyls and Junlin Wang and Sanjay Subramanian and Matt Gardner and Sameer Singh},
                      Booktitle = {Empirical Methods in Natural Language Processing},                            
                      Year = {2019},
                      Title = {AllenNLP Interpret: A Framework for Explaining Predictions of NLP Models}}

Projects

Link to some of my for fun projects