INDEX

Explanations

call the police

The neuron fires on words that signal direct commands or requests (imperative mood), such as “go,” “please,” “meet,” or “deliver.”

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

 навіть

-1.86

pantalon

-1.54

 даже

-1.52

 même

-1.52

 varier

-1.51

伤感

-1.45

DAFTAR

-1.41

 nécessairement

-1.40

 innehåller

-1.40

sepatu

-1.40

POSITIVE LOGITS

 then

2.06

 before

1.67

if

1.65

we

1.52

 there

1.46

 since

1.46

how

1.45

 what

1.42

 that

1.41

 some

1.40

Activations Density 0.024%