When Every Choice Is Correct: A Causal Study of Numerical Preference in Language Models
cs.CL
Submitted: 2026-11-04
Updated: 2026-11-04
Terminology
Sources
- BiasBusters: Uncovering and Mitigating Tool Selection Bias in Large Language Models
- Deterministic or probabilistic? The psychology of LLMs as random number generators
- The Illusion of Stochasticity in LLMs
- Qwen3 Technical Report
- Causal Mediation Analysis for Interpreting Neural NLP: The Case of Gender Bias
- Towards Best Practices of Activation Patching in Language Models: Metrics and Methods
- Is This the Subspace You Are Looking for? An Interpretability Illusion for Subspace Activation Patching
- The Hydra Effect: Emergent Self-repair in Language Model Computations
- Perturbation Probing: A Two-Pass-per-Prompt Diagnostic for FFN Behavioral Circuits in Aligned LLMs
- Leverage Is Not Reach: A Control-Window Law for Single-Neuron Steering in Language Models
- Benford's Curse: Tracing Digit Bias to Numerical Hallucination in LLMs
- 3, 7, 14, 17, 42: Instruction Tuning Installs and Sharpens Number-Choice Heuristics in Language Models
- Does the Model Use the Feature? Separating Steering from Mechanism in LLMs
- Base Models Beat Aligned Models at Randomness and Creativity
- AtP*: An efficient and scalable method for localizing LLM behaviour to components
- Interpretability in the Wild: a Circuit for Indirect Object Identification in GPT-2 small
Related papers
- ALARM: Audio-Language Alignment for Reasoning Models
- ThinkSum: Probabilistic reasoning over sets using large language models
- Coherence boosting: When your pretrained language model is not paying enough attention
- From Facts to Insights: A Persona-Driven Dual Memory Framework and Dataset for Role-Playing Agents
- Improving General Role-Playing Agents via Psychology-Grounded Reasoning and Role-Aware Policy Optimization
- The Art of Asking: Multilingual Prompt Optimization for Synthetic Data