INDEX

Explanations

closing parentheses followed by various words

The neuron fires on foreign‐language tokens (non-English words or phrases in various scripts).

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

你先

-2.20

<bos>

-2.14

 cráneo

-2.09

你别

-2.06

و

-1.88

靣

-1.81

 siento

-1.77

 pijn

-1.76

,''

-1.75

ではまた

-1.73

POSITIVE LOGITS

are

2.67

ของ

1.98

Są

1.74

);

1.69

 leggera

1.68

🩹

1.68

νομα

1.66

ICOLON

1.64

 named

1.63

魮

1.62

Activations Density 0.006%