INDEX

Explanations

auditors and divisors

The neuron flags technical or agent‐noun terms ending in an “-or/-isor” suffix.

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

 terutama

-2.50

with

-2.48

(!__

-2.47



-2.39

 overdue

-2.34

ellido

-2.34



-2.34

iNdEx

-2.27

龊

-2.25

あります

-2.23

POSITIVE LOGITS

</h3>

2.84

2.66

le

2.66

</i>

2.61

</h1>

2.58

si

2.48

)।

2.44

聼

2.42

</sub>

2.39

↵

2.38

Activations Density 0.003%