INDEX

Explanations

The neuron strongly responds to the token “mix,” i.e. occurrences of the substring “mix.”

mix Adhesive, USS Portsmouth, Two small

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

of

-2.05

on

-1.93

for

-1.77

-1.74

</h3>

-1.64

an

-1.63

at

-1.54

any

-1.51

 this

-1.50

you

-1.45

POSITIVE LOGITS

がありますが

1.82

がなく

1.64

 accentu

1.61

Lịch

1.60

<bos>

1.59

Ofic

1.48

でも

1.46

では

1.45

Ответить

1.45

تفسير

1.44

Activations Density 0.000%