INDEX

Explanations

The neuron fires on single‐word action verbs in their base form (like “raise,” “transform,” “cancel,” “remove,” “draw,” “use,” “defines”) that typically denote operations or commands.

actions and their consequences

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

gorm

-1.05

of

-1.03

する

-0.99

 Such

-0.98

collants

-0.95

 عبار

-0.94

mişti

-0.93

 должен

-0.91

Granted

-0.91

ฤษ

-0.91

POSITIVE LOGITS

 them

1.48

its

1.34

 được

1.23

 zich

1.16

 themselves

1.13

ถูก

1.11

笑了

1.09

 worden

1.09

 itself

1.06

 their

1.05

Activations Density 0.149%