INDEX

Explanations

confirming veracity

The neuron detects occurrences of evidence- or proof-related terms (e.g. “proving,” “proves,” “proven”) and conditional markers like “if.”

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits



0.60

Lat

0.56

possibly

0.55

}}=$

0.55

Lat

0.55

可能會

0.55

 brochures

0.54

 frosted

0.53

'$,

0.53

жер

0.53

POSITIVE LOGITS

 proving

1.52

证明

1.50

 prove

1.48

 proves

1.45

 Prove

1.42

證明

1.34

Prove

1.33

 proved

1.32

 подтвер

1.30

 साबित

1.27

Activations Density 0.276%