INDEX

Explanations

introduces clauses and examples

The neuron fires most strongly on words that appear as the first significant (non‐stop) word of a new segment or sentence.

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

 разных

-1.61

เพราะ

-1.26

各类

-1.23

걍

-1.20

 variously

-1.20

 經典

-1.18

 stavu

-1.17

 różne

-1.16

 různých

-1.16

 różnych

-1.16

POSITIVE LOGITS

the

1.66

1.30

将会

1.20

1.16

的模样

1.15

あく

1.14

1.13

1.11

 idéia

1.11

的笑容

1.11

Activations Density 0.061%