INDEX

Explanations

code delimiters

The neuron flags JSON‐style structural tokens—brackets, braces, and quoted key names—marking the syntax of JSON data.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

</h2>

-2.08

-1.82

 also

-1.81

by

-1.67

or

-1.60

for

-1.59

-1.58

</strong>

-1.56

It

-1.55

POSITIVE LOGITS

the

1.86

 cinto

1.77

łgorzata

1.73

 adesi

1.67

 necesar

1.64

 munc

1.62

 speciale

1.59

 έχουν

1.59

 futuras

1.58

 dita

1.57

Activations Density 0.001%