INDEX

Explanations

studies show

the neuron detects references to scientific studies, research findings, and related statistical evidence.

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

 with

-1.53

茄子

-1.39

焙

-1.38

 while

-1.35

Dí

-1.31

̓

-1.30

 "}";

-1.30

😷

-1.27

meneu

-1.27

 Архивировано

-1.26

POSITIVE LOGITS

 kwamba

1.75

 abbildung

1.64

 effectuer

1.57

踮

1.48

↵↵↵↵↵↵↵↵

1.41

 publicidade

1.40

 studies

1.34

 udało

1.34

 пять

1.33

↵↵↵

1.32

Activations Density 0.108%