INDEX

Explanations

Fahrenheit, MOSFET, deciduous, Pasadena

The neuron activates on numeric tokens—especially decimal‐point values (e.g. “116.5,” “102.5,” “88.5”)—i.e. it’s detecting numbers with fractional parts.

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

</h3>

-3.16

</i>

-2.72

isation

-2.59

?"

-2.42

th

-2.38

 When

-2.33

ال

-2.25

toplankton

-2.23

</strong>

-2.22

This

-2.16

POSITIVE LOGITS

↵↵

2.81

 ſame

2.58

ühungen

2.53

are

2.50

鼍

2.44

쐈

2.39

 leaſt

2.39

 núm

2.36

嗩

2.36



2.30

Activations Density 0.016%