INDEX

Explanations

purchaser, acquirer, pancreas, JSP

The neuron fires on tokens that begin with the letter “p.”

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

</strong>

-3.58

毟

-2.75

葙

-2.70

曏

-2.58

 palav

-2.58



-2.58

枼

-2.44

nya

-2.42

 rocha

-2.41

 teto

-2.38

POSITIVE LOGITS

</td>

2.63

2.31

2.14

 undue

2.13

↵↵

2.08

荽

2.03

begin

2.02

 вернулся

1.99

1.95

end

1.95

Activations Density 0.003%