INDEX

Explanations

Russian and other languages

The neuron fires strongly on comparative degree markers and related words (e.g. rather, than, far, least, first two) used to signal comparisons.

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

-1.59

 These

-1.55

 those

-1.53

 Those

-1.38

 Regardless

-1.33

 what

-1.32

TWO

-1.29

 these

-1.28

 such

-1.28

Those

-1.27

POSITIVE LOGITS

nadier

1.45

aconda

1.45

 новых

1.41

髕

1.37

睒

1.35

 новые

1.31

Autoritní

1.30

嫆

1.27

襌

1.27

 нового

1.27

Activations Density 0.155%