INDEX

Explanations

degree and quality modifiers

The neuron specializes in spotting degree words or intensifiers (adverbs/adjectives that convey emphasis or strength of degree).

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

などが

-1.41

 hien

-1.31

 maten

-1.23

を活用

-1.20

がなく

-1.19

 alip

-1.17

 fuis

-1.15

があり

-1.15

 Grüßen

-1.13

 lastly

-1.12

POSITIVE LOGITS

It

1.60

づけ

1.59

ができます

1.45

The

1.37

 http

1.29

1.25

1.24

鏽

1.21

 suoi

1.19

 οποία

1.19

Activations Density 0.032%