How High Is 0.6? Floors, Ceilings, and Headroom in Interpretability Probing
cs.CL, cs.AI
Submitted: 2026-10-06
Updated: 2026-10-06
Terminology
Sources
- What learning algorithm is in-context learning? Investigations with linear models
- Understanding intermediate layers using linear classifier probes
- Designing a Dashboard for Transparency and Control of Conversational AI
- Interpretability Illusions in the Generalization of Simplified Models
- TabPFN-3: Technical Report
- Emergent World Representations: Exploring a Sequence Model Trained on a Synthetic Task
- The Geometry of Truth: Emergent Linear Structure in Large Language Model Representations of True/False Datasets
- Transformers Can Do Bayesian Inference
- Solving Empirical Bayes via Transformers
- No Training Required: Exploring Random Encoders for Sentence Classification
- A Theory of Usable Information Under Computational Constraints
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