INDEX

Explanations

high and advanced modifiers

This neuron detects adjectives and related terms that emphasize high degree, quality, or advanced/superior properties (e.g. “high,” “advanced,” “sophisticated,” “highest,” etc.).

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

 That

-1.55

 This

-1.52

 There

-1.50

 artful

-1.45

<bos>

-1.42

 amenaz

-1.39

𐑂

-1.36

 With

-1.35

towane

-1.35

To

-1.34

POSITIVE LOGITS

或者

1.84

͊

1.41

 fiabilidad

1.39

 considérer

1.39

 dekoratif

1.38

鷽

1.38

 incontri

1.36

 iniziale

1.35

俅

1.35

 junto

1.34

Activations Density 0.028%