INDEX

Explanations

blame for

The neuron strongly activates on the word “for,” especially when it appears in contexts of assigning blame or responsibility.

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

躥

-1.77

摃

-1.59

omiast

-1.52

ियम

-1.52

 postulates

-1.46

𝐅

-1.46

 appartenant

-1.45

☎

-1.44

vetrain

-1.42

шении

-1.41

POSITIVE LOGITS

for

1.95

1.75

↵

1.49

to

1.48

 with

1.47

หรือ

1.45

"../

1.42

1.41

rendre

1.40

at

1.39

Activations Density 0.010%