INDEX

Explanations

The neuron never activates—it doesn’t detect any pattern.

code or structured data delimiters

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

of

-0.95

 since

-0.94

relenting

-0.92

 throughout

-0.90

}=\

-0.90

asjon

-0.88

个

-0.87

 Anlage

-0.84

 abgerufen

-0.84

october

-0.84

POSITIVE LOGITS

tischer

1.06

uristic

1.04

帑

1.02

 mít

0.94

 doigt

0.91

Modelos

0.91

 sentier

0.90

fasterxml

0.88

 dismant

0.88

Masa

0.88

Activations Density 0.001%