INDEX

Explanations

This neuron detects directive or advisory language—i.e. imperative commands and recommendations (e.g. “Don’t go here,” “Buy this bag,” “Check your credit‐card orders”).

don't go this way

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

clubs

-1.05

🤓

-1.01

gulls

-1.00

righted

-1.00

colours

-0.99

 finland

-0.98

☺☺

-0.97

 paraf

-0.97

👋

-0.97

mrt

-0.96

POSITIVE LOGITS

if

1.52

 this

1.30

 anyone

1.20

 если

1.20

 jeśli

1.16

 here

1.16

 yourselves

1.02

you

1.00

 tästä

0.99

 اگر

0.98

Activations Density 0.006%