INDEX

Explanations

speech attribution after quote

This neuron activates on dialogue attribution tags—speaker-attribution verbs like “said,” “replied,” “echoed,” etc., often accompanied by the surrounding punctuation.

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

बाबत

0.69

 rahul

0.59

 singh

0.58

 dotyczą

0.58

 gdyż

0.56

 Aditya

0.55

While

0.54

 concernant

0.54

 surpasses

0.53

 while

0.52

POSITIVE LOGITS

َم

0.74

任意の

0.64

 හෝ

0.62

 EXPL

0.62

Methoxy

0.61

 ...)

0.61

--");

0.59

দ্‌

0.59

?')

0.59

治療

0.58

Activations Density 0.005%