INDEX

Explanations

following a colon or specific words

The main thing this neuron does is detect literal numeric values (including hex color codes and other number tokens).

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

ремонт

-0.81

粱

-0.80

ських

-0.79

 después

-0.78

ད

-0.77

র

-0.77

ีก

-0.75

<bos>

-0.75

oherty

-0.75

nonce

-0.75

POSITIVE LOGITS

 default

0.92

Anyway

0.87

Cohen

0.80

Maharashtra

0.79

 Cohen

0.79

Blessed

0.76

いま

0.75

默认

0.75

 creciendo

0.74

 trov

0.72

Activations Density 0.013%