INDEX

Explanations

here to help

The neuron fires strongly on service-offer or reassurance phrases—especially “we are here to…”/“we’re here to…” constructions.

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

笵

-1.73

觧

-1.67

 wciąż

-1.60

autogui

-1.60

衚

-1.58

靄

-1.57

ساخت

-1.55

 sentarse

-1.53

 dopo

-1.52

Etimología

-1.52

POSITIVE LOGITS

to

2.84

in

1.68



1.57

</strong>

1.49

–

1.48

despite

1.45

—

1.42



1.41

later

1.39

and

1.38

Activations Density 0.013%