INDEX

Explanations

The neuron strongly activates on content words (open‐class words like nouns, verbs, and adjectives) and ignores function words.

style and trends

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 choisissez

-1.01

 indescri

-0.97

 entrando

-0.93

Страна

-0.93

 Segel

-0.92

 evidentemente

-0.88

sre

-0.87

ፊ

-0.87

 tämä

-0.85

 Fische

-0.85

POSITIVE LOGITS

ulence

1.18

ough

1.13

 what

1.11

€™

1.05

our

1.05

lique

1.04

 Cymru

0.99

 }_{\

0.98

OUGH

0.97

<",

0.97

Activations Density 0.097%