INDEX

Explanations

way

The neuron strongly activates on high–information content words—especially concrete nouns, proper names, time‐ or count‐based terms, and descriptive adjectives.

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

在前

0.85

在

0.84

ябрь

0.79

在本

0.72

жению

0.71

在该

0.70

 inherent

0.70

क्तियों

0.70

ใน

0.69

 Чтобы

0.69

POSITIVE LOGITS

 وفي

1.21

 guise

1.20

 веке

1.09

 וב

1.08

 формате

1.07

 Weise

1.05

 midst

1.05

 режиме

1.02

 своём

1.01

 accordance

1.00

Activations Density 1.392%