INDEX

Explanations

program code

The neuron responds to unusually rare or specific words (high-IDF 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

на

0.78

 sfida

0.76

 chạm

0.75

ем

0.74

micron

0.74

showthread

0.72

िस

0.71

nip

0.71

뭐

0.70

į

0.70

POSITIVE LOGITS

ificateur

0.89

antaranya

0.74

 الماء

0.73

 faciliter

0.73

 usersRouter

0.73

 NEED

0.72



0.72

TPU

0.71

ATOR

0.70

ordine

0.70

Activations Density 0.000%