INDEX

Explanations

evidence and proof

The neuron detects references to empirical scientific terms—words like data, evidence, observations, experiments, and related concepts.

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

-1.84

ByDefault

-1.54

 when

-1.46

if

-1.39

had

-1.38

 werd

-1.37

:['

-1.34

 jika

-1.34

„

-1.33

愉

-1.30

POSITIVE LOGITS

 that

1.69

れてる

1.45

曖昧さ回避

1.30

 الفلسط

1.25

する方法

1.23

outros

1.22

 に関する

1.22

 verstor

1.21

知り

1.20

メルカリ

1.20

Activations Density 0.226%