INDEX

Explanations

something else

The main thing this neuron does is detect occurrences of the word “else” (as in “something else,” “somebody else’s,” “someone else,” etc.).

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

-1.55

䨻

-1.52

 dokter

-1.51

akses

-1.46

颏

-1.42

econom

-1.41

 marqués

-1.40

 JOHN

-1.39

ers

-1.39

ﮧ

-1.38

POSITIVE LOGITS

 than

2.48

 then

1.94

」、「

1.89

 equally

1.76

tawesome

1.75

 similarly

1.73

 also

1.71

 затем

1.69

 indicate

1.62

 likewise

1.59

Activations Density 0.013%