INDEX

Explanations

games, matches, or films

The neuron fires on longer, content‐heavy words—especially technical or scientific terms and key verbs—typically found in formal, data-driven or methodological contexts.

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

 affich

-1.40

ఝ

-1.38

officially

-1.37

 vendu

-1.34

naturally

-1.32

directly

-1.30

purchased

-1.30

 vider

-1.30

acquired

-1.29

 acheter

-1.28

POSITIVE LOGITS

 with

1.88

 when

1.82

or

1.66

if

1.65

at

1.63

 before

1.62

not

1.61

for

1.56

 only

1.55

to

1.51

Activations Density 0.081%