INDEX

Explanations

The neuron flags multiword proper‐noun phrases and formal titles—like names of organizations, awards, medical conditions, or specific entity names.

concepts and associated organizations

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

ίας

-1.32

and

-1.27

 necessarie

-1.26

 berle

-1.26

 your

-1.22

 orejas

-1.19

 tomba

-1.19

-1.18

 things

-1.15

 perlas

-1.15

POSITIVE LOGITS

🢒

1.16

<bos>

1.15

of

1.14

 legislación

1.13

 contextLoads

1.11

𝟻

1.08

 哪些

1.08

hysema

1.07

 tři

1.05

 čtyř

1.03

Activations Density 0.077%