INDEX

Explanations

study method approach

The neuron fires on high-level methodological or study-type nouns (e.g. “studies,” “analysis,” “approach,” “method”) that describe the research or service being presented.

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

and

-2.48

as

-1.53

künfte

-1.37

 kjer

-1.34

Appellants

-1.33

lioma

-1.29

 مختلفة

-1.29

 kommit

-1.28

 including

-1.26

valuer

-1.26

POSITIVE LOGITS

1.70

1.61

1.53

1.49

1.41

 baseado

1.39

."

1.38

 لأنه

1.36

 prace

1.30

Activations Density 0.328%