INDEX

Explanations

hero

The neuron fires on mentions of “hero” (and its morphological variants like “heroes,” “heroine,” etc.).

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

-2.55

閙

-1.96

...)

-1.94

儈

-1.84

мо

-1.82

..."

-1.82

ஂ

-1.81

='

-1.79

 produkter

-1.79

 affectionate

-1.78

POSITIVE LOGITS

↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵

2.13

 soste

2.08

将他

2.02

↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵

1.99

1.98

筈

1.98

↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵

1.98

跟她

1.97

 овощи

1.95

↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵

1.95

Activations Density 0.009%