INDEX

Explanations

fake and false information

The neuron is specialized to spot mentions of “fake” (especially in the phrase “fake news”) or closely related falsehood cues.

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

 ‘‘

-1.53

-1.48

他有

-1.47

 словно

-1.46

 sué

-1.39

-1.38

 крайне

-1.34

すでに

-1.33

-1.31

replyreport

-1.28

POSITIVE LOGITS

</

1.67

 ['./

1.51

 False

1.49

 cercano

1.45

 time

1.45

 Fake

1.44

 decreto

1.43

 "/")

1.39

 significativo

1.38

=()

1.38

Activations Density 0.027%