INDEX

Explanations

specific word following prompts

The neuron primarily detects numeric tokens written with decimal points (i.e. floating‐point numbers).

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

for

-2.27

 There

-1.60

 corriendo

-1.56

 Like

-1.56

But

-1.55

 Because

-1.55

 These

-1.51

řízení

-1.51

but

-1.50

 This

-1.50

POSITIVE LOGITS

you

2.25

鹇

1.84

进入了

1.75

 publicly

1.55

𓐍

1.52

 ウィメンズ

1.48

骉

1.47

 augusti

1.46

the

1.45

艄

1.45

Activations Density 0.171%