INDEX

Explanations

html declaration namespaces

The neuron strongly fires on four-digit year tokens (e.g. “1999”) in XML/HTML namespace or DOCTYPE declarations.

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

了一

-1.88

It

-1.73

);

-1.60

for

-1.55

has

-1.41

-1.39

 coincident

-1.38

又是一

-1.38

}$.

-1.36

POSITIVE LOGITS

窣

1.59

 favoriser

1.54

 that

1.54

鑁

1.52

ांकि

1.51

ﺲ

1.50

蜘

1.48

 teller

1.48

asanjo

1.46

THAT

1.45

Activations Density 0.004%