INDEX

Explanations

features, properties, and changes

The neuron detects high-level abstract nouns used to label attributes, categories, or findings (e.g. “features,” “factors,” “properties,” “characteristics,” “requirements,” “effects,” etc.) in technical or scientific text.

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

one

-1.41

 predominant

-1.30

？！

-1.28

 Without

-1.26

พอ

-1.25

 vieron

-1.25

autorité

-1.25

Without

-1.24

躂

-1.23

 without

-1.22

POSITIVE LOGITS

 that

2.75

 which

1.44



1.38

you

1.37

 including

1.35

圌

1.34

 quotidi

1.34

僦

1.31

 UNIQUE

1.28



1.27

Activations Density 0.336%