INDEX

Explanations

address, types, patient, old

The neuron flags semantically rich content words—i.e. non-stop words carrying substantial meaning (like “different,” “microprocessor,” “attacker,” etc.).

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



-1.45

provides

-1.34

妩

-1.30

allows

-1.23

ſy

-1.21

 lisäksi

-1.20

becomes

-1.19

殽

-1.19

鸶

-1.18

 aikana

-1.17

POSITIVE LOGITS

to

2.22

it

1.83

 with

1.68

 like

1.66

as

1.55

not

1.52

 because

1.52

 until

1.51

for

1.51

 their

1.50

Activations Density 0.044%