INDEX

Explanations

code brackets and symbols

The neuron primarily activates on numeric literals (number constants) in the 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

the

-1.67

 obtuvo

-1.64

-1.60

ing

-1.53

 praça

-1.44

 informiert

-1.38

过去

-1.33

 twice

-1.30

 classed

-1.30

 being

-1.28

POSITIVE LOGITS

:],

1.89

škin

1.72

揠

1.71

}}>

1.63

⇨

1.63

iência

1.60

égique

1.58

nji

1.57

鞴

1.56

詒

1.55

Activations Density 0.038%