INDEX

Explanations

for you

The neuron predominantly fires on content‐bearing words—main nouns and verbs that carry the core meaning of a 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

薄

-0.72

ranslated

-0.69

chada

-0.67

uristic

-0.67

 conversar

-0.65

△

-0.65

 ライブ

-0.64

baan

-0.64

trina

-0.64

 Kult

-0.64

POSITIVE LOGITS

 עבור

2.16

 بالنسبة

1.54

 hộ

1.33

 sizin

1.24

åt

1.24

 számára

1.23

forme

1.21

 मेरे

1.20

 benim

1.20

 bagi

1.19

Activations Density 0.034%