INDEX

Explanations

often, especially, sometimes, almost

The neuron responds most strongly to unusually long tokens—i.e. longer, multi‐syllable words.

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

 that

-2.16

 with

-1.88

and

-1.52

 using

-1.49

one

-1.44

-1.37

-1.36

美国的

-1.29

pography

-1.27

小說

-1.27

POSITIVE LOGITS

sometimes

1.84

often

1.64

especially

1.62

 soprattutto

1.56

since

1.54

 çünkü

1.51

versaire

1.48

 than

1.43

almost

1.41

vacy

1.40

Activations Density 0.212%