INDEX

Explanations

open, pending, outstanding

This neuron fires strongly on words that signal unresolved items or issues (e.g., “outstanding,” “pending,” “open,” “overdue,” “questions,” “challenges”).

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

-1.64

-1.54

-1.51

-1.41

drav

-1.39

菒

-1.38

-1.37

 verbind

-1.34

⡪

-1.33

-1.30

POSITIVE LOGITS

),

1.43

 motivadoras

1.41

Hozzá

1.41

回戦

1.37

Hogyan

1.35

 painfully

1.34

 будь

1.28

 Unterschied

1.25

 básicas

1.25

 (†

1.24

Activations Density 0.018%