INDEX

Explanations

is expected

The neuron flags rare or information-dense content words (i.e. uncommon, key tokens) in the text.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 dioxane

1.02

 teatral

0.90

ﺠ

0.89

 parete

0.85

 cellSize

0.84

 ikinci

0.83

 tenga

0.83

 scegliere

0.83

 richiest

0.83

 sputum

0.82

POSITIVE LOGITS

the

0.82

ides

0.76

ensing

0.76

rik

0.75

icking

0.73

ocl

0.72

 Bright

0.71

 рождения

0.71

ុក

0.70

Дру

0.70

Activations Density 0.000%