INDEX

Explanations

phrases starting with "from"

the neuron responds most strongly to frequent function words (like “a,” “the,” “up,” “however,” etc.) and other high‐utility English terms, i.e. common stop words.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 thing

-1.20

-1.15

 boletín

-1.09

-1.07

-1.06

官员

-1.05

 things

-1.01

 на

-1.00

POSITIVE LOGITS

 standpoint

1.42

 outset

1.41

}>;

1.30

 afar

1.23

뻗

1.22

aktivi

1.14

🛐

1.13

⃝

1.09

pecabe

1.07

 大学

1.06

Activations Density 0.377%