INDEX

Explanations

parentheses with content

This neuron specifically fires on numeric literals (and related arithmetic symbols) in code.

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

を

-1.27

<h2>

-1.03

 провести

-0.98

Only

-0.98

 that

-0.98

 ONLY

-0.97

Ở

-0.96

 Only

-0.96

四川

-0.95

そして

-0.94

POSITIVE LOGITS

 ouvertes

1.07

へ向

1.07

 disponíveis

0.95

 abiertas

0.94

 тыся

0.94

 ouvert

0.92

期の

0.91

 ouverts

0.88

 qualifi

0.88

__':

0.88

Activations Density 0.005%