INDEX

Explanations

capable, ready, compatible, enabled, friendly

The neuron activates on descriptive adjectives that flag product features or capabilities—especially ones ending in –able, –ible, –ed, or –y, like “ready,” “capable,” “enabled,” “compatible,” “programmable,” or “friendly.”

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

it

-1.25

❋

-1.20

 okrąg

-1.15

ję

-1.13

 sahiptir

-1.11

In

-1.10

ocide

-1.09

 rękaw

-1.09

 enviada

-1.07

 Оно

-1.07

POSITIVE LOGITS

without

1.34

Current

1.28

corbic

1.27

but

1.25

corporated

1.23

νη

1.23

perger

1.22

1.21

又能

1.21

phalt

1.21

Activations Density 0.048%