INDEX

Explanations

E-mails, League, car, linked

The neuron detects single uppercase letters immediately followed by a hyphen (e.g. the “X-” prefix in terms like “O-linked,” “P-glycoprotein,” “E-selectin,” etc.).

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

-1.80

-1.73

-1.58

 decorativo

-1.52

or

-1.52

-1.50

或

-1.48

 cruda

-1.43

 changed

-1.41

‘

-1.41

POSITIVE LOGITS

だけでなく

1.89

 también

1.80

 them

1.77

螃

1.76

鲩

1.70

 její

1.68

 และ

1.66

睃

1.63

 comprensión

1.59

tiéndose

1.59

Activations Density 0.079%