INDEX

Explanations

topics followed by descriptions

This neuron consistently fires on the central topic word of a passage—typically the key subject or headline noun appearing early in each paragraph.

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

 this

-2.41

 what

-2.11

 when

-2.09

any

-1.97

or

-1.83

if

-1.73

 just

-1.70

 their

-1.63

 only

-1.56

 every

-1.55

POSITIVE LOGITS

 денег

1.34

also

1.23

often

1.23

was

1.23

⢓

1.23

፱

1.22

а

1.21

常常

1.21

usually

1.20

 havde

1.19

Activations Density 0.001%