Research map / Probabilistic and scientific modeling / Variational and generative sampling
Omitted Variable Bias in Language Models Under Distribution Shift
Victoria Lin, Louis-Philippe Morency, Eli Ben-Michael
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| optimize a worst-case generalization bound instead of standard DR objective | linear model on LIWC features · Amazon (semi-synthetic) | test performance | Improved |
| increase assumed omitted-variable strength product up to about 0.8 (ablation) | worst-case objective · EmoBank | test MSE vs DR: peak gain around 0.8 | Improved |
| increase assumed omitted-variable strength product up to about 0.06 (ablation) | worst-case objective · Hate Speech | test BCE vs DR: peak gain around 0.06 | Improved |
| over-increase the assumed sensitivity beyond the peak (ablation) | worst-case objective · EmoBank and Hate Speech | test performance | Hurt |
| account for omitted variables via worst-case bound rather than DR estimate | GPT-4.1 evaluation · MATH-Perturb | estimated target performance | Improved |
Closest papers by meaning
- Large language models suffer from a curse of ambiguity (NeurIPS 2026 · similarity 0.89)
- A Probabilistic Perspective on Unlearning and Alignment for Large Language Models (ICLR 2025 · similarity 0.88)
- Understanding Multimodal LLMs Under Distribution Shifts: An Information-Theoretic Approach (ICML 2025 · similarity 0.88)
- Biases Mitigation and Expressiveness Preservation in Language Models: A Comprehensive Pipeline (Student Abstract) (AAAI 2024 · similarity 0.88)
- Leveraging robust optimization for llm alignment under distribution shifts (NeurIPS 2025 · similarity 0.88)
- Paloma: A Benchmark for Evaluating Language Model Fit (NeurIPS 2024 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
