INDEX

Explanations

superlatives and subjective states

The neuron fires on abstract, discourse‐structuring nouns common in academic or argumentative prose (e.g. “model,” “reasons,” “evidence,” “question,” “sequence,” “requirement,” “effect,” “number,” “structure,” etc.).

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

Nullable

0.65

 পাইয়া

0.61

certified

0.60

<unused26>

0.58

 গিয়েছিলাম

0.58

日的

0.58

SIC

0.57

camouflage

0.57

SWE

0.56

новения

0.56

POSITIVE LOGITS

of

0.91

iest

0.88

niest

0.75

is

0.72

នៃការ

0.68

igste

0.66

liest

0.64

 outweighs

0.64

ของการ

0.63

ziest

0.63

Activations Density 0.000%