INDEX

Explanations

connecting to LLMs or data

The neuron responds strongly to moderately common English content words (nouns, adjectives, pronouns, and adverbs) rather than to rare or highly specialized terms.

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

 colored

0.80

 pelajar

0.76

 днем

0.75

 coloured

0.69

 Colored

0.68

 எப்போதும்

0.66

 ओवर

0.66

يں

0.65

 Сара

0.65

 ঘণ্টায়

0.65

POSITIVE LOGITS

るので

0.81

 yeux

0.80

나

0.76

Maj

0.75

ας

0.74

0.73

の頃

0.73

Mn

0.72

 Masa

0.72

Activations Density 0.000%