Jack Hessel

Research Scientist @ AI2

contact: jackh@allenai.org

CV (as of 9/2022).

I am on github, twitter, google scholar, and semantic scholar.

I am a Research Scientist at AI2. Previously, I earned a PhD in Computer Science at Cornell University. These days, I'm most interested in improving human-AI collaboration. This includes:

• developing methods to align model behavior with human intent;
• probing the limits of scaling to identify where models can most benefit from human intervention;
• and expanding models with new modalities (e.g., vision+language) for a more complete view of the world.

Some recent preprints

• Do Androids Laugh at Electric Sheep? Humor "Understanding" Benchmarks from The New Yorker Caption Contest
  Jack Hessel, Ana Marasović, Jena D. Hwang, Lillian Lee, Jeff Da, Rowan Zellers, Robert Mankoff, Yejin Choi
  The New Yorker Dataset is now available at capcon.dev
• Is Reinforcement Learning (Not) for Natural Language Processing?: Benchmarks, Baselines, and Building Blocks for Natural Language Policy Optimization
  Rajkumar Ramamurthy, Prithviraj Ammanabrolu, Kianté Brantley, Jack Hessel, Rafet Sifa, Christian Bauckhage, Hannaneh Hajishirzi, Yejin Choi
  Check out the RL4LMs library!

Publications (in reverse chronological order)

• Quark: Controllable Text Generation with Reinforced Unlearning
  Ximing Lu, Sean Welleck*, Jack Hessel*, Liwei Jiang, Lianhui Qin, Peter West, Prithviraj Ammanabrolu, Yejin Choi.
  NeurIPS 2022 (oral)
  code+camera ready coming soon!
• The Abduction of Sherlock Holmes: A Dataset for Visual Abductive Reasoning.
  Jack Hessel*, Jena D Hwang*, Jae Sung Park, Rowan Zellers, Chandra Bhagavatula, Anna Rohrbach, Kate Saenko, and Yejin Choi.
  ECCV 2022 (oral)
  dataset/code/leaderboard; press: venturebeat
• Reframing Human-AI Collaboration for Generating Free-Text Explanations.
  Sarah Wiegreffe, Jack Hessel, Swabha Swayamdipta, Mark Riedl, and Yejin Choi.
  NAACL 2022
  code
• Connecting the Dots between Audio and Text without Parallel Data through Visual Knowledge Transfer.
  Yanpeng Zhao, Jack Hessel, Youngjae Yu, Ximing Lu, Rowan Zellers, and Yejin Choi.
  NAACL 2022
  code
• Symbolic Knowledge Distillation: from General Language Models to Commonsense Models
  Peter West, Chandra Bhagavatula, Jack Hessel, Jena D. Hwang, Liwei Jiang, Ronan Le Bras, Ximing Lu, Sean Welleck, and Yejin Choi
  NAACL 2022
  code, explainer video
• MERLOT Reserve: Neural Script Knowledge through Sound, Language, and Vision
  Rowan Zellers, Jiasen Lu, Ximing Lu, Youngjae Yu, Yanpeng Zhao, Mohammadreza Salehi, Aditya Kusupati, Jack Hessel, Ali Farhadi, Yejin Choi
  CVPR 2022 (oral)
  code, project page, Press: Venturebeat, Press: GeekWire, with interview from Rowan!
  
• MERLOT: Multimodal Neural Script Knowledge Models
  Rowan Zellers*, Ximing Lu*, Jack Hessel*, Youngjae Yu, Jae Sung Park, Jize Cao, Ali Farhadi, and Yejin Choi.
  NeurIPS 2021 (oral)
  code, project page, Press: The Batch, Press: Venturebeat
  
• CLIPScore: A Reference-free Evaluation Metric for Image Captioning
  Jack Hessel, Ari Holtzman, Maxwell Forbes, Ronan Le Bras, and Yejin Choi
  EMNLP 2021
  code, talk
  
• How effective is BERT without word ordering? Implications for language understanding and data privacy.
  Jack Hessel and Alexandra Schofield
  ACL 2021
  code, talk
  
• Does my multimodal model learn cross-modal interactions? It’s harder to tell than you might think!
  Jack Hessel and Lillian Lee
  EMNLP 2020
  code, talk
  
• Beyond Instructional Videos: Probing for More Diverse Visual-Textual Grounding on YouTube
  Jack Hessel, Zhenhai Zhu, Bo Pang, and Radu Soricut
  EMNLP 2020
  data, talk
  
• Domain-Specific Lexical Grounding in Noisy Visual-Textual Documents
  Gregory Yauney, Jack Hessel, and David Mimno
  EMNLP 2020
  code
  
• Learning from Multimodal Web Data
  PhD Thesis
  Cornell University 2020
  
• Unsupervised Discovery of Multimodal Links in Multi-image, Multi-sentence Documents
  Jack Hessel, Lillian Lee, and David Mimno
  EMNLP 2019
  code/data, poster, project page
• A Case Study on Combining ASR and Visual Features for Generating Instructional Video Captions
  Jack Hessel, Bo Pang, Zhenhai Zhu, and Radu Soricut
  CoNLL 2019
  talk slides
• Something's Brewing! Early Prediction of Controversy-causing Posts from Discussion Features
  Jack Hessel and Lillian Lee
  NAACL 2019
  data
• Quantifying the Visual Concreteness of Words and Topics in Multimodal Datasets
  Jack Hessel, David Mimno, and Lillian Lee
  NAACL 2018
  code/data
• Cats and Captions vs. Creators and the Clock: Comparing Multimodal Content to Context in Predicting Relative Popularity
  Jack Hessel, Lillian Lee, David Mimno
  WWW 2017
  code/data, slides, replication
• Aligning Images and Text in a Digital Library
  Jack Hessel and David Mimno
  Computer Vision in Digital Humanities Workshop at DH 2017 (extended abstract)
• Science, AskScience and BadScience: On the Coexistience of Highly Related Communities
  Jack Hessel, Chenhao Tan, and Lillian Lee
  ICWSM 2016
  data, slides, replication
• What do Vegans do in their Spare Time? Latent Interest Detection in Multi-Community Networks
  Jack Hessel, Alexandra Schofield, Lillian Lee, David Mimno
  Workshop on Networks in the Social and Information Sciences at NeurIPS 2015
  project page
• Image Representations and New Domains in Neural Image Captioning
  Jack Hessel, Nicolas Savva, and Kimberly J. Wilber
  Workshop on Vision and Language at EMNLP 2015
  slides
• Using Reproductive Altruism to Evolve Multicellularity in Digital Organisms
  Jack Hessel and Sherri Goings
  ECAL 2013

Work Experience

• I was a postdoc on the Mosaic team at AI2 from October 2020 through June 2021.
• I spent the summer of 2019 working with Google's natural language understanding team.
• I was a visiting faculty member at Carleton College in Spring, 2019; I taught two classes: Natural Language Processing and Discrete Math.
• I spent the summer of 2018 working with Google's natural language understanding team.
• I spent the summer of 2017 working with Facebook's core data science team.
• I spent the summer of 2016 working with the Twitter Cortex team in NYC.

Invited Talks

Service/Guest Lectures/Other Activites

Other fun projects

• I wrote a TreeLSTM in tensorflow 2; this is a neural network whose topology changes based on each input example.
• I wrote a factorization machine layer in pytorch; for speed reasons, the forward and backward passes are written in cython.
• I implemented Monroe et al.'s "Fightin' Words" algorithm for robustly comparing word frequencies in two corpora. This implementation has been used in several publications, e.g., this and this
• As part of an NSF REU, I contributed to the implementation of a really (really) fast SVM solver in Kilian Weinberger's lab at Washington University, St. Louis. [now at Cornell!] see Parallel Support Vector Machines in Practice by Tyree, S., Gardner, J. R., Weinberger, K. Q., Agrawal, K., & Tran, J. (2014).
• "A Comparative Analysis of Popular Phylogenetic Reconstruction Algorithms." Undergraduate thesis project/best paper award at MICS 2014: joint work with Evan Albright, Nao Hiranuma, Cody Wang, and Sherri Goings.

More?

I grew up in beautiful Portola Valley, California. I earned a B.A. from Carleton College in 2014, studying computer science and mathematics/statistics. During my time in Northfield, I played ice hockey, and hosted a radio show. I even returned to Carleton briefly in 2019, this time, as a visiting faculty member. I'm a die hard San Jose Sharks fan, avid consumer (and very occasionally a producer) of electronic music, and, an amateur lockpick. During graduate school at Cornell, I was a member of Stewart Little Coop, a community of 15 people, I played ice hockey in the Ithaca Hockey Association (and, during summer internships in CA, in the San Jose Adult Hockey League).