INDEX

Explanations

comparisons using than

The neuron consistently detects occurrences of the word “less.”

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

 ženy

-1.65

 kimse

-1.48

 Selasa

-1.46

and

-1.45

kün

-1.39

趼

-1.38

नलोड

-1.31

 Tanpa

-1.30

dengar

-1.30

-1.28

POSITIVE LOGITS

the

1.80

on

1.55

its

1.45

 efectu

1.43

1.42

“

1.39

、「

1.38

 sicherzustellen

1.38

−/−

1.37

1.34

Activations Density 0.018%