INDEX

Explanations

intensifiers followed by descriptors

The neuron detects degree-intensifying adverbs (e.g. “very,” Chinese 很, Russian много) that modify adjectives or verbs.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

tront

-2.22

fillment

-1.91

aneity

-1.87

 唯美

-1.86

yship

-1.83

März

-1.83

tifiable

-1.82

 排骨

-1.82

 tutur

-1.81

 特写

-1.81

POSITIVE LOGITS

1.63

<strong>

1.61

,'

1.47

1.45

1.44

1.43

 rare

1.41

有一些

1.41

•

1.39

1.38

Activations Density 0.004%