INDEX

Explanations

of

The neuron strongly activates on attribution and reporting verbs (e.g. “said,” “added,” “tells,” “explains”) that introduce quotations or speaker attributions.

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

 floppy

0.70

eax

0.66

もしくは

0.64

 داخل

0.61

≤

0.60

 verdi

0.60

 toroidal

0.59

ޤ

0.59

 hacía

0.59

 elongated

0.59

POSITIVE LOGITS

：“

1.43

，“

1.38

，「

1.30

："

1.29

：「

1.28

 "...

1.28

"[

1.21

「

1.20

:“

1.20

"'

1.16

Activations Density 0.200%