INDEX

Explanations

headers like H or specific tokens like 7

The neuron strongly activates on any token containing digits (numeric literals or identifiers with numbers).

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

着一个

-2.13

-2.08

on

-1.94

if

-1.82

So

-1.75

崋

-1.70

and

-1.69

 compi

-1.66

});

-1.65

}{

-1.65

POSITIVE LOGITS

 hacían

1.88

珰

1.87

’

1.77

jawab

1.75

 zimowe

1.73

CurtirCurtir

1.73

 llevaron

1.72

 espectadores

1.72

 Warszawie

1.70

⁧

1.70

Activations Density 0.041%