Neurons exhibit concept-conditioned activation ranges forming Gaussian-like distributions with minimal overlap, and range-based interventions via NeuronLens outperform neuron-level masking in targeted manipulation with reduced collateral effects.
Title resolution pending
3 Pith papers cite this work, alongside 961 external citations. Polarity classification is still indexing.
3
Pith papers citing it
961
external citations · OpenAlex
verdicts
UNVERDICTED 3representative citing papers
Empirical audit of LAION-2B-en and LAION-2B-multi finds overrepresentation of young adults, White people, and males plus stereotypical emotion associations across two attribute classifiers.
Empirical comparison shows gradient-based explanations for GNN node similarities are actionable, consistent, and retain effects when sparsified, unlike mutual information explanations.
citing papers explorer
-
Neurons Speak in Ranges: Breaking Free from Discrete Neuronal Attribution
Neurons exhibit concept-conditioned activation ranges forming Gaussian-like distributions with minimal overlap, and range-based interventions via NeuronLens outperform neuron-level masking in targeted manipulation with reduced collateral effects.
-
Unmasking LAION-5B: Age, Gender, Race, and Emotion Biases in Large-Scale Image Datasets
Empirical audit of LAION-2B-en and LAION-2B-multi finds overrepresentation of young adults, White people, and males plus stereotypical emotion associations across two attribute classifiers.
-
Explaining Graph Neural Networks for Node Similarity on Graphs
Empirical comparison shows gradient-based explanations for GNN node similarities are actionable, consistent, and retain effects when sparsified, unlike mutual information explanations.