INDEX

Explanations

weighing, worse, degree

The neuron fires on evaluative language—words used to make judgments or comparisons (e.g. “weigh,” “weighing,” “worse,” “off-base,” etc.).

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

}\

-1.66

 grises

-1.65

》，

-1.58

”،

-1.56

婓

-1.55

AMAZING

-1.52

Beispiel

-1.52

 Berlín

-1.52

}=\

-1.51

"):

-1.51

POSITIVE LOGITS

 such

2.06

 that

1.73

if

1.55

 solcher

1.52

If

1.48

our

1.47

 according

1.46

 every

1.41

 each

1.41

 podía

1.34

Activations Density 0.095%