INDEX

Explanations

закон, данных, слов, окно

The neuron is detecting contiguous Cyrillic (Russian) text segments.

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

vermectin

-1.17

believe

-1.13

ėjimo

-1.12

*/,

-1.08

ėjimas

-1.05

 इसकी

-1.04

 hinauf

-1.03

 которое

-1.02

 ہیں۔

-1.02

bout

-1.01

POSITIVE LOGITS

not

1.89

is

1.77

 also

1.72

 still

1.69

 always

1.55

 never

1.53

 doesn

1.45

 considered

1.41

 supposed

1.39

 definitely

1.35

Activations Density 0.017%