INDEX

Explanations

dates and identifiers

The neuron strongly responds to structural/documentation tokens—things like section or metadata markers (titles, headers, dates, subtitles, etc.) rather than normal sentence text.

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

taba

-0.88

֙

-0.85

橢

-0.82

})->

-0.80

all

-0.79

Ң

-0.79

 amarilla

-0.75

 acoged

-0.74

setFirstName

-0.72

plusplus

-0.71

POSITIVE LOGITS

<b>

1.48

<i>

1.05

 votes

1.00

Vote

0.94

Voting

0.91

ਗ਼

0.91

Word

0.91

VOTE

0.90

————

0.90

Author

0.87

Activations Density 0.002%