INDEX

Explanations

pairing

This neuron detects the presence of quotation marks (including single, double, and triple quote delimiters).

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

そこは

-1.64

Also

-1.55

 éta

-1.52

なかな

-1.52

 frasco

-1.48

eamiento

-1.46

 peines

-1.43

 multicolore

-1.42

Similarly

-1.41

追記

-1.41

POSITIVE LOGITS

 semua

1.82

↵

1.77

すべての

1.51

%");

1.48

 have

1.46

 setiap

1.45

()");

1.45

%).

1.41

\%$

1.38

all

1.38

Activations Density 0.056%