INDEX

Explanations

underlying philosophy

The neuron strongly activates on content‐heavy descriptive words—i.e. formal, multi‐syllable adjectives and adverbs used to qualify or emphasize (e.g. “sufficient,” “inherent,” “necessary,” “wider,” “further,” “given”).

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 thing

0.80

0.78

 smoker

0.75

 cantar

0.74

 stuff

0.74

సారి

0.74

 haunts

0.72

一下子

0.72

 folder

0.72

 dingen

0.72

POSITIVE LOGITS

 reimag

0.89

MyToolkit

0.83

ásticos

0.81

ámicas

0.80

́ng

0.79

 활용

0.79

ASTM

0.78

ייתה

0.77

 স্বায়ত্ত

0.77

 interconnected

0.76

Activations Density 0.505%