INDEX

Explanations

also, too

The neuron strongly activates on non-Latin text characters—especially CJK (Chinese/Japanese) ideographs.

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

and

-2.83

-2.80

of

-2.17

-2.11

 Because

-2.05

 With

-2.05

 From

-1.91

 karena

-1.88

もしくは

-1.87

bzw

-1.86

POSITIVE LOGITS

 abbildung

1.95

🪆

1.94



1.89

 sintética

1.85

 ersche

1.81

躥

1.81

 comunista

1.78

 herbes

1.78

 gustaMe

1.77

⠪

1.76

Activations Density 0.001%