Fragility Is Not Reducible to Hardness: Conditional Causal Difficulty for LLMs
cs.LG
Submitted: 2026-11-01
Updated: 2026-11-01
Code: https://github.com/upwork/Conditional_Causal_
Terminology
Sources
- Contrastive representations of high-dimensional, structured treatments
- Learning Compact Representations of LLM Abilities via Item Response Theory
- Optimizing Prompts for Large Language Models: A Causal Approach
- Fragile Reasoning: A Mechanistic Analysis of LLM Sensitivity to Meaning-Preserving Perturbations
- Measuring Mathematical Problem Solving With the MATH Dataset
- Adaptive Testing for LLM Evaluation: A Psychometric Alternative to Static Benchmarks
- From Residuals to Reasons: LLM-Guided Mechanism Inference from Tabular Data
- Reliable and Efficient Amortized Model-based Evaluation
- Compared to What? Baselines and Metrics for Counterfactual Prompting
Related papers
- FlexMoE: Learning One-for-All Multi-Budget Expert Pruning for Flexible MoE Deployment
- OptiMind: Improving Small Language Models for Optimization Modeling through Formulation Guidance and Multi-Turn Learning
- Staying in the Sweet Spot: Boost Reasoning Evolution via Responsive Rollout Difficulty Manipulation
- Examining the Impact of Data and Training Methods on Emergent and Subliminal Misalignment
- Rethinking Black-Box On-Policy Distillation: A Rubric-based Analysis
- (Mis)generalization of Helpful-Only Fine-tuning