INDEX

Explanations

particularly

This neuron detects first‐person narrative cues—pronouns like “I” (and related personal expressions such as “can’t,” “tell you,” “how,” etc.) indicating the speaker’s own perspective.

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

ariConfig

-1.00

 ORCH

-0.97

 ostavi

-0.97

 lourdes

-0.96

 pérd

-0.93

翆

-0.93

 miljö

-0.93

 this

-0.91

 regler

-0.91

ább

-0.90

POSITIVE LOGITS

 επίσης

1.19

Also

1.07

 particularly

1.02

 esimerkiksi

0.93

 forthcoming

0.93

 Particularly

0.92

 także

0.91

Among

0.91

 manteniendo

0.89

Two

0.88

Activations Density 0.093%