INDEX

Explanations

increasing trend

The neuron is looking for the repeated intensifier phrase “more and more.”

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

鸰

-2.16

But

-1.95

 muchos

-1.91

 irmão

-1.84

hilangan

-1.82

 granja

-1.79

It

-1.78

is

-1.74

 одним

-1.73

 ながら

-1.71

POSITIVE LOGITS

of

2.47

 them

1.76

 familières

1.73

—

1.66

gador

1.64

ور

1.61

вание

1.60

 these

1.59

ğına

1.57

lijden

1.55

Activations Density 0.004%