INDEX

Explanations

returning values or program states

The neuron consistently activates on the “return” keyword 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

-1.55

 disgu

-1.27

-1.15

seit

-1.14

 большую

-1.13

 мою

-1.13

 многих

-1.12

葑

-1.12

 inget

-1.09

ーツ

-1.07

POSITIVE LOGITS

 their

1.49

__":

1.46

.“

1.40

 mostly

1.18

目前

1.16

ſelves

1.16

 barato

1.15

 feels

1.13

 nuevo

1.12

!”

1.11

Activations Density 0.001%