Spontaneous Context Restoration: How Language Models Recover from Corrupted Inputs

arXiv:2609.35475 · cs.CL, cs.AI · Submitted 2026-09-28 · Read on arXiv

cs.CL, cs.AI

Submitted: 2026-09-28

Updated: 2026-09-28

Terminology

• Ablation — 39× in this paper · explained in 1 episode
• Transformer — 25× in this paper · explained in 2 episodes
• Robustness — 23× in this paper · explained in 5 episodes
• Accuracy — 22× in this paper · explained in 1 episode
• AUC — 21× in this paper · explained in 2 episodes
• Cosine Similarity — 16× in this paper · explained in 3 episodes
• Alignment — 10× in this paper · explained in 1 episode
• Linear probe — 10× in this paper · explained in 2 episodes
• Activation Patching — 8× in this paper · explained in 2 episodes
• Fine-Tuning — 7× in this paper · explained in 5 episodes
• Verification — 6× in this paper · explained in 2 episodes
• Trace — 4× in this paper · explained in 2 episodes
• Transformer Models — 4× in this paper · explained in 2 episodes
• Benchmark — 2× in this paper · explained in 2 episodes
• Interpretability — 2× in this paper · explained in 15 episodes
• Adam — 1× in this paper · explained in 2 episodes
• Adaptation — 1× in this paper · explained in 1 episode
• Chain-of-Thought Prompting — 1× in this paper · explained in 2 episodes
• Consistency — 1× in this paper · explained in 3 episodes
• Core — 1× in this paper · explained in 2 episodes
• Decode — 1× in this paper · explained in 2 episodes
• F1 Score — 1× in this paper · explained in 3 episodes
• Generalization — 1× in this paper · explained in 13 episodes
• In-Context Learning — 1× in this paper · explained in 4 episodes
• Iterative Refinement — 1× in this paper · explained in 3 episodes
• Quantization — 1× in this paper · explained in 12 episodes
• Transformer Architecture — 1× in this paper · explained in 2 episodes

Sources

• Cutting Through the Noise: Boosting LLM Performance on Math Word Problems
• Eliciting Latent Predictions from Transformers with the Tuned Lens
• Exploring LLM Reasoning Through Controlled Prompt Variations
• Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challenge
• Not All Language Model Features Are One-Dimensionally Linear
• Localizing Model Behavior with Path Patching
• How to use and interpret activation patching
• The Remarkable Robustness of LLMs: Stages of Inference?
• Improving the Robustness of Large Language Models for Code Tasks via Fine-tuning with Perturbed Data
• The Geometry of Truth: Emergent Linear Structure in Large Language Model Representations of True/False Datasets
• Right for the Wrong Reasons: Diagnosing Syntactic Heuristics in Natural Language Inference
• The Hydra Effect: Emergent Self-repair in Language Model Computations
• How Context Affects Language Models' Factual Predictions
• SQuAD: 100,000+ Questions for Machine Comprehension of Text
• Explorations of Self-Repair in Language Models
• Robustness of Large Language Models to Perturbations in Text
• A Mechanistic Interpretation of Arithmetic Reasoning in Language Models using Causal Mediation Analysis
• Interpretability in the Wild: a Circuit for Indirect Object Identification in GPT-2 small
• Interpreting Arithmetic Mechanism in Large Language Models through Comparative Neuron Analysis
• Towards Best Practices of Activation Patching in Language Models: Metrics and Methods

Related papers

• Exploring Solution Divergence and Its Effect on Large Language Model Problem Solving
• Ishigaki-IDS-Bench: A Benchmark for Generating Information Delivery Specification from BIM Information Requirements
• Subliminal Steering: Stronger Encoding of Hidden Signals
• MedStruct-S: A Benchmark for Key Discovery, Key-Conditioned QA and Semi-Structured Extraction from OCR Clinical Reports
• The End of Transformers? On Challenging Attention and the Rise of Sub-Quadratic Architectures
• Untangling the Mechanisms of Misleading Context in Medical Question Answering