INDEX

Explanations

setting to default

The neuron flags numeric literal values (especially when they appear in assignments or “set … to” 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

 jambe

-1.88

횃

-1.70

斯克

-1.68

Breathe

-1.66

eee

-1.65

\}}

-1.62

 cuchara

-1.61

蓠

-1.60

ództ

-1.59

]")]

-1.58

POSITIVE LOGITS

in

2.03

the

1.88

as

1.73

to

1.69

 emphatically

1.66

1.60

 need

1.55

赑

1.55

 since

1.51

呌

1.50

Activations Density 0.062%