INDEX

Explanations

quick followed by specific nouns

The main thing this neuron does is detect the token “Quick.”

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

-2.22

‘

-2.19

-2.16

).

-2.11

but

-1.91

と言われる

-1.90

風の

-1.88

on

-1.84

-1.81

Let

-1.80

POSITIVE LOGITS

They

2.59

Setiap

2.36

𓍯

2.34

 itſelf

2.30

 émotion

2.28

勠

2.25

 kebaya

2.25

 milano

2.22

 jogja

2.22

 parachoque

2.19

Activations Density 0.008%