INDEX

Explanations

counting out of

This neuron specializes in detecting the word “out” as used in the common “X out of Y” pattern (e.g. “4 out of 5”).

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

”

-2.23

</h1>

-1.97

 (‘

-1.93

,’

-1.91

…”

-1.80

 (“

-1.76

‘

-1.74

:”

-1.73

ி

-1.73

しつつ

-1.68

POSITIVE LOGITS

of

2.17

鹀

1.95

Karakteristik

1.95

 Ketika

1.84

 verhind

1.83

无语

1.82

 kurzem

1.79

Ketika

1.72

をチェック

1.71

approximately

1.70

Activations Density 0.005%