INDEX

Explanations

code delimiters and function calls

The neuron is keyed to programming‐style identifiers—especially function or method names (e.g. cursor, DoEvents, Props) rather than normal English words.

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

 into

-1.20

 then

-1.08

}:{

-1.04

$=$

-1.02

 james

-1.00

 than

-0.99

!!!!!!!!!

-0.98

并

-0.98

!!!!!!!

-0.97

 następnie

-0.95

POSITIVE LOGITS

 boca

1.07

 citar

1.04

 mysterious

1.02

औ

1.02

 establishing

1.02

虽说

1.00

there

0.98

Cinta

0.98

的历史

0.97

 sombr

0.96

Activations Density 0.013%