Profiles2020-2025
Jacob Steinhardt
University of California
18Papers
8Last author
31Unique co-authors
0First author
0Solo
First and last author come from byline position, which some venues assign alphabetically.
Recent papers18
- Adversaries Can Misuse Combinations of Safe ModelsAuthor 3 of 4ICML 2025
- Interpreting the Second-Order Effects of Neurons in CLIPAuthor 3 of 4ICLR 2025
- Iterative Label Refinement Matters More than Preference Optimization under Weak SupervisionAuthor 3 of 3ICLR 2025
- Uncovering Gaps in How Humans and LLMs Interpret Subjective LanguageAuthor 3 of 4ICLR 2025
- Interpreting CLIP's Image Representation via Text-Based DecompositionAuthor 3 of 4ICLR 2024 · 179 citations
- Approaching Human-Level Forecasting with Language ModelsAuthor 5 of 5NeurIPS 2024 · 142 citations
- Overthinking the Truth: Understanding how Language Models Process False DemonstrationsAuthor 3 of 4ICLR 2024 · 83 citations
- Describing Differences in Image Sets with Natural LanguageAuthor 7 of 9CVPR 2024
- Jailbroken: How Does LLM Safety Training Fail?Author 4 of 4NeurIPS 2023 · 2,230 citations
- Mass-Producing Failures of Multimodal Systems with Language ModelsAuthor 3 of 4NeurIPS 2023 · 54 citations
- Interpretability in the Wild: a Circuit for Indirect Object Identification in GPT-2 SmallAuthor 5 of 5ICLR 2023 · 50 citations
- Capturing Failures of Large Language Models via Human Cognitive BiasesAuthor 3 of 3NeurIPS 2022 · 154 citations
- Predicting Out-of-Distribution Error with the Projection NormAuthor 5 of 5ICML 2022 · 51 citations
- Grounding Representation Similarity Through Statistical TestingAuthor 3 of 3NeurIPS 2021 · 88 citations
- Learning Equilibria in Matching Markets from Bandit FeedbackAuthor 3 of 4NeurIPS 2021 · 52 citations
- Are Larger Pretrained Language Models Uniformly Better? Comparing Performance at the Instance LevelAuthor 4 of 4ACL Findings 2021 · 24 citations
- Rethinking Bias-Variance Trade-off for Generalization of Neural NetworksAuthor 4 of 5ICML 2020 · 219 citations
- Identifying Statistical Bias in Dataset ReplicationAuthor 5 of 6ICML 2020 · 55 citations
