INDEX

Explanations

while loop conditions

The neuron is looking for loop‐control expressions—especially “while” conditions and their comparison or update tokens.

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

 then

-2.33

 already

-2.25

 just

-1.98

now

-1.89

if

-1.84

 want

-1.77

 since

-1.66

 уже

-1.59

 currently

-1.48

as

-1.46

POSITIVE LOGITS

ícias

1.68

yeah

1.47



1.43

iation

1.41

岽

1.38

沨

1.37

voer

1.37

€™

1.36

 errichtet

1.36

tively

1.34

Activations Density 0.010%