INDEX

Explanations

beginning and end phrases

The neuron fires most strongly on words marking narrative or section boundaries—specifically “end” and “beginning.”

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

 from

-2.25

in

-2.17

at

-1.92

and

-1.63

 viviendas

-1.54

 összefoglaló

-1.37

 принад

-1.37

 inmiddels

-1.36

嵙

-1.34

 myśli

-1.33

POSITIVE LOGITS

 químicos

1.60

缡

1.57

 expérience

1.53

 использование

1.52

鼙

1.48

 lequel

1.47

]--;

1.45

ственный

1.44

 échelle

1.43

 нескольких

1.41

Activations Density 0.019%