INDEX

Explanations

this object reference

The neuron activates on occurrences of the keyword “this” in code snippets.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

of

-2.72

 which

-1.59

 will

-1.57

do

-1.47

if

-1.46

-1.41

 each

-1.38

 love

-1.37

 also

-1.36

it

-1.33

POSITIVE LOGITS

うーん

1.79

goes

1.64

ӄ

1.59

 исключительно

1.54

ihe

1.52

despite

1.49

guarantee

1.47



1.45

knows

1.45

どうやら

1.42

Activations Density 0.005%