INDEX

Explanations

states or actions

The neuron detects technical process verbs—especially passive‐voice constructions (forms of “be” plus past participles) commonly used to describe steps or operations.

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

!」

-1.67

</h6>

-1.45

\"></

-1.33

!'

-1.29

seled

-1.23

veloped

-1.22

-”

-1.20

.!

-1.20

xique

-1.20

€

-1.20

POSITIVE LOGITS

or

1.60

 Также

1.40

慟

1.35

 When

1.34

 Also

1.30

 তার

1.30

 Although

1.29

 obdob

1.28

 Actually

1.27

 Compared

1.23

Activations Density 0.336%