INDEX

Explanations

explanations of reasons why

The neuron detects discourse markers indicating explanations or justifications—especially the word “reasons” (and variants like “why”) that introduce explanatory clauses.

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

1.17

age

1.11

Ba

1.07

1.05

灿

1.02



1.01

\(

0.98

Back

0.97

Bo

0.96

POSITIVE LOGITS

Reasons

1.23

 reasons

1.18

 Natives

1.14

 reason

1.13

why

1.09

 السبب

1.09

ciones

1.08

ną

1.08

 razón

1.06

理由は

1.06

Activations Density 0.029%