INDEX

Explanations

patterns in sentence structure

The neuron fires on individual content‐bearing nouns (key topic words) rather than common function words.

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

 another

-1.34

 OGSÅ

-1.31

 superfí

-1.27

 reembols

-1.26

 livelli

-1.22

béco

-1.21

 rather

-1.20

 perché

-1.20

 retrovisor

-1.18

 agarre

-1.17

POSITIVE LOGITS

 आपके

1.32

 tillegg

1.19

<bos>

1.18

 Nacionales

1.16

 сегодня

1.13

 will

1.10

 توانید

1.09

),

1.09

erey

1.08

 будете

1.07

Activations Density 0.222%