INDEX

Explanations

place/look/heel/overhead/Rose/K/search followed by word

The neuron strongly responds to numbered, step-by-step exercise instructions (e.g. “Step 1: Lie down…”, “Step 2: Inhale and raise…”).

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

 vinaigre

-1.44

：“……”

-1.35

 that

-1.34

 чехол

-1.32

ecie

-1.30

to

-1.30

 stof

-1.27

 бампер

-1.25

tismo

-1.23

Inhoud

-1.23

POSITIVE LOGITS

 your

1.67

 with

1.37

 Nedir

1.35

Dziękuję

1.32

 katholischen

1.31

 tumbuhan

1.30

Pozdrawiam

1.28

GRACIAS

1.27

stdarg

1.22

 државе

1.21

Activations Density 0.012%