INDEX

Explanations

tested positive

The neuron detects mentions of positive test results (e.g. “tested positive for …” or “[X] positive for …”).

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

</h5>

-1.72

</h1>

-1.70

):

-1.59

-1.57

↵↵

-1.52

ḥ

-1.52

如果

-1.49

-1.44

他在

-1.44

')

-1.43

POSITIVE LOGITS

 they

1.51

 encontraban

1.49

 heridos

1.48

#,

1.45

 handful

1.43

おわりに

1.41

resents

1.34

 vile

1.34

 pname

1.33

、【

1.30

Activations Density 0.008%