INDEX

Explanations

variable names and loop counters

The neuron is triggered by the standalone lowercase token “c” (e.g. as a single-letter variable or flag).

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

but

-1.38

Literat

-1.29

 ....

-1.26

澳门

-1.23

 ......

-1.23

Quite

-1.22

 before

-1.19

 interessieren

-1.18

 .....

-1.16

idades

-1.14

POSITIVE LOGITS

 incredi

1.56

 said

1.38

1.36

];

1.33

 GUIDELINES

1.30

を選

1.30

閖

1.27

 indispens

1.20

 ADVISORY

1.20

</strong>

1.19

Activations Density 0.082%