When Probing Accuracy Saturates, Fragility Resolves: A Complementary Metric for LLM Pre-Training Analysis

arXiv:2606.11375 · cs.CL, cs.AI, cs.LG · Submitted 2026-06-09 · Read on arXiv

cs.CL, cs.AI, cs.LG

Submitted: 2026-06-09

Updated: 2026-08-25

Code: https://github.com/deepsteer/deepsteer

Terminology

• Accuracy — 133× in this paper · explained in 1 episode
• Robustness — 14× in this paper · explained in 5 episodes
• Transformer — 10× in this paper · explained in 2 episodes
• Linear Probing — 8× in this paper · explained in 2 episodes
• Generalization — 7× in this paper · explained in 13 episodes
• Redundancy — 7× in this paper · explained in 2 episodes
• Fine-Tuning — 6× in this paper · explained in 5 episodes
• Linear probe — 6× in this paper · explained in 2 episodes
• Phase Transitions — 5× in this paper · explained in 2 episodes
• Ablation — 4× in this paper · explained in 1 episode
• Core — 4× in this paper · explained in 2 episodes
• Interpretability — 4× in this paper · explained in 15 episodes
• APEX — 3× in this paper · explained in 1 episode
• Benchmark — 3× in this paper · explained in 2 episodes
• Decode — 3× in this paper · explained in 2 episodes
• Adaptation — 2× in this paper · explained in 1 episode
• Empirical Methods in Natural Language Processing — 2× in this paper
• LoRA Fine-Tuning — 2× in this paper · explained in 3 episodes
• Adam — 1× in this paper · explained in 2 episodes
• Alignment — 1× in this paper · explained in 1 episode
• Anisotropy — 1× in this paper · explained in 3 episodes
• Consistency — 1× in this paper · explained in 3 episodes
• In International Conference on Learning Representations — 1× in this paper
• In-Context Learning — 1× in this paper · explained in 4 episodes
• Meeting of the Association for Computational Linguistics — 1× in this paper
• Representation Engineering — 1× in this paper · explained in 2 episodes
• Scaffolding — 1× in this paper · explained in 3 episodes
• Verification — 1× in this paper · explained in 2 episodes
• alpha — 1× in this paper · explained in 1 episode

Sources

• Understanding intermediate layers using linear classifier probes
• Refusal in Language Models Is Mediated by a Single Direction
• Pythia: A Suite for Analyzing Large Language Models Across Training and Scaling
• Walking Noise: On Layer-Specific Robustness of Neural Architectures against Noisy Computations and Associated Characteristic Learning Dynamics
• OLMo: Accelerating the Science of Language Models
• Designing and Interpreting Probes with Control Tasks
• Locating and Editing Factual Associations in GPT
• Progress measures for grokking via mechanistic interpretability
• 2 OLMo 2 Furious
• Information-Theoretic Probing for Linguistic Structure
• Grokking: Generalization Beyond Overfitting on Small Algorithmic Datasets
• APEX: Probing Neural Networks via Activation Perturbation
• Information-Theoretic Probing with Minimum Description Length
• Representation Engineering: A Top-Down Approach to AI Transparency

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