Transformer hidden states for numbers exhibit anti-scalar variability (alpha ≈ -0.19) rather than the constant coefficient of variation found in biological magnitude systems.
Weber’s law in transformer magnitude representations: Efficient coding, representational geometry, and psychophysical laws in language models, 2026
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
years
2026 2verdicts
UNVERDICTED 2representative citing papers
Causal localization via attribution and patching identifies a temporal preference subgraph in mid-to-upper layers of Qwen3-4B-Instruct-2507, with time-horizon geometry in the residual stream and initial evidence for steering-vector control.
citing papers explorer
-
Same Geometry, Opposite Noise: Transformer Magnitude Representations Lack Scalar Variability
Transformer hidden states for numbers exhibit anti-scalar variability (alpha ≈ -0.19) rather than the constant coefficient of variation found in biological magnitude systems.
-
Temporal Preference Concepts and their Functions in a Large Language Model
Causal localization via attribution and patching identifies a temporal preference subgraph in mid-to-upper layers of Qwen3-4B-Instruct-2507, with time-horizon geometry in the residual stream and initial evidence for steering-vector control.