INDEX

Explanations

how much / as much / so much

The neuron fires on degree-expressing adverbs (like “as,” “so,” “not” in comparative or intensifier constructions).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

Better

-0.85

better

-0.81

 flere

-0.79

صدار

-0.78

Welche

-0.76

Ϲ

-0.75

Сер

-0.75

Weitere

-0.74

させて頂

-0.74

ങ്ങള

-0.74

POSITIVE LOGITS

 much

7.47

much

4.91

Much

4.53

 Much

4.25

 MUCH

3.34

 mycket

2.91

 paljon

2.69

 mucho

2.48

 viel

2.38

มาก

2.16

Activations Density 0.044%