INDEX

Explanations

is or are followed by description

The neuron is picking out passive‐voice markers—i.e. occurrences of “to be” (is/are/was/etc.) and the preposition “by” in passive constructions.

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

-2.16

-1.96

-1.93

-1.91

那些

-1.76

-1.75

 terke

-1.74

\|_{

-1.74

可以

-1.72

POSITIVE LOGITS

and

2.31

on

2.05

boneca

1.90

for

1.87

 away

1.83

 known

1.81

 órgãos

1.80

 núcleos

1.80

cakup

1.79

nissan

1.77

Activations Density 0.054%