Safety Pretraining: Toward the Next Generation of Safe AI
Pratyush Maini*, Sachin Goyal*, Dylan Sam*, Alex Robey, Yash Savani, Yiding Jiang, Andy Zou, Zachary C. Lipton, J. Zico Kolter
[pdf, website]
Analyzing Similarity Metrics for Data Selection for Language Model Pretraining
Dylan Sam, Ayan Chakrabarti, Afshin Rostamizadeh, Srikumar Ramalingam, Gui Citovsky, Sanjiv Kumar
[pdf]
Predicting the Performance of Black-box LLMs thorugh Self-Queries
Dylan Sam, Marc Finzi, and J. Zico Kolter
ICML Next Generation of AI Safety Workshop, 2024
[pdf, code]
Open Challenges in Applying Language Models to 3D Molecular Data
Nate Gruver, Marc Finzi, Dylan Sam, J. Zico Kolter, Ben Athiwaratkun, and Andrew Gordon Wilson
Under Review
Finetuning CLIP to Reason about Pairwise Differences
Dylan Sam, Devin Willmott, Joao D. Semedo, and J. Zico Kolter
[pdf, code]
Understanding Prompt Engineering Does Not Require Rethinking Generalization
Victor Akinawade, Yiding Jiang, Dylan Sam, and J. Zico Kolter
International Conference on Learning Representations (ICLR), 2024
ICML Sampling and Optimization in Discrete Spaces, 2023 (Outstanding Paper)
[pdf]
Auditing Fairness under Unobserved Confounders
Yewon Byun, Dylan Sam, Michael Oberst, Zachary C. Lipton, and Bryan Wilder.
International Conference on Artificial Intelligence and Statistics (AISTATS), 2024
[pdf, code]
Computing Low-Entropy Couplings for Large-Support Distributions
Samuel Sokota, Dylan Sam, Christian Schroeder de Witt, Spencer Compton, Jakob Nicolaus Foerster, and J. Zico Kolter
Conference on Uncertainty in Artificial Intelligence (UAI), 2024
[pdf, code]
Bayesian Neural Networks with Domain Knowledge Priors
Dylan Sam*, Rattana Pukdee*, Daniel P. Jeong, Yewon Byun, and J. Zico Kolter
ICML Knowledge and Logical Reasoning Workshop, 2023(Oral)
[pdf, code]
Learning with Explanation Constraints
Rattana Pukdee*, Dylan Sam*, J. Zico Kolter, Maria-Florina Balcan, and Pradeep Ravikumar
Neural Information Processing Systems (NeurIPS), 2023
[pdf]
Label Propagation with Weak Supervision
Rattana Pukdee*, Dylan Sam*, Maria-Florina Balcan, and Pradeep Ravikumar
International Conference on Learning Representations (ICLR), 2023
[pdf,
code]
Losses over Labels: Weakly Supervised Learning via Direct Loss Construction
Dylan Sam and J. Zico Kolter
AAAI Conference on Artificial Intelligence, 2023
[pdf,
code]
Improving self-supervised representation learning via sequential adversarial masking
Dylan Sam, Min Bai, Tristan McKinney, and Li Erran Li
NeurIPS 2022 Workshop: Self-Supervised Learning - Theory and Practice, 2022
[pdf]
Adversarial Multi Class Learning under Weak Supervision with Performance Guarantees
Alessio Mazzetto*, Cyrus Cousins*, Dylan Sam, Stephen H Bach, and Eli Upfal
International Conference on Machine Learning (ICML), 2021
[pdf,
code]
Semi-Supervised Aggregation of Dependent Weak Supervision Sources with Performance Guarantees
Alessio Mazzetto, Dylan Sam, Andrew Park, Eli Upfal, and Stephen H Bach
International Conference on Artificial Intelligence and Statistics (AISTATS), 2021
[pdf,
code]
Learning from Dependent Weak Supervision Sources
Dylan Sam
Undergraduate Honors Thesis, 2021
[pdf]
Automated Data Accountability for the Mars Science Laboratory
Ryan Alimo, Dylan Sam, Dounia Lakhmiri, Brian Kahovec, and Dariush Divsalar
IEEE Aerospace Conference, 2021
[pdf,
code]
Hierarchical Clustering Analysis of Spectral Fingerprints for Cheminformatics
Dylan Sam and Brenda M Rubenstein
NeurIPS Machine Learning for Molecules Workshop, 2020
[pdf]