INDEX

Explanations

special characters followed by keywords

The neuron activates on numeric constants and arithmetic offsets (e.g. literal numbers and expressions like “+135”, “-82.5”, “91.5”) in code.

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

for

-2.59

in

-2.44

 with

-2.30

 while

-2.08

not

-2.06

 during

-2.06

 before

-1.96

-1.94

 without

-1.92

 only

-1.91

POSITIVE LOGITS

 этих

2.02

全都

1.80

 ซึ่ง

1.78

 myriad

1.77

 différents

1.77

 kurze

1.75

 åren

1.75

 cette

1.75

搪

1.73

 keinerlei

1.71

Activations Density 0.006%