INDEX

Explanations

saying something specific

The neuron fires on first‐person self-references, especially tokens like “I” and nearby verbs (e.g. “say,” “said”), marking personal statements.

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

DLJ

1.48

이었다

1.41

يا

1.39

ADOR

1.38

 sulfides

1.38

 fistule

1.31

 transients

1.29

DebugInformation

1.29

 hypersurfaces

1.29

 inodes

1.28

POSITIVE LOGITS

tion

1.82

liness

1.61

tte

1.49

driven

1.48

 goodbye

1.48

scopic

1.45

tional

1.45

м

1.41

то

1.40

tur

1.40

Activations Density 0.068%