INDEX

Explanations

why this is so

The neuron primarily detects the intensifier “so.”

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

镲

-2.38

ఽ

-2.28

强行

-2.19

妧

-2.17

 новую

-2.16

 substan

-2.13



-2.13

鹀

-2.09

 .........

-2.05

 apropri

-2.05

POSITIVE LOGITS

in

3.11

 with

2.63

 Those

2.44

for

2.42

at

2.34

 after

2.33

。「

2.31

 when

2.30

。“

2.20

on

2.17

Activations Density 0.011%