INDEX

Explanations

Yes, following statements

The neuron detects question sentences by strongly activating on the question‐mark token “?”.

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

and

-1.72

 also

-1.28

 what

-1.26

 with

-1.25

 también

-1.24

 from

-1.20

how

-1.16

 which

-1.16

expectedResult

-1.12

to

-1.09

POSITIVE LOGITS

you

2.44

 your

1.93

大丈夫です

1.39

you

1.25

your

1.22

那你

1.19

カッコいい

1.17

áš

1.16

了你

1.14

そこは

1.14

Activations Density 0.078%