INDEX

Explanations

not or no

This neuron detects negation words or constructions (e.g. “not,” “does not”) indicating a negative statement.

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

 แต่

-1.67

}=

-1.54

of

-1.40

(!__

-1.33

–

-1.30

but

-1.28

他的

-1.27

}.

-1.25

gettimeofday

-1.24

 jedoch

-1.23

POSITIVE LOGITS

 evitare

1.42

1.40

ikkert

1.32

 roues

1.31

是为了

1.27

 câble

1.23

 möglich

1.20

 seguire

1.19

Noter

1.19

 любой

1.17

Activations Density 0.017%