INDEX

Explanations

out of

The neuron fires strongly on the token “out,” i.e. it detects occurrences of the word “out.”

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 uart

1.07

 autorisé

1.06

 humaine

1.02

 aporta

1.02

 invités

1.01

ımın

1.00

ه‌ای

0.96

はこちら

0.95

 antérieure

0.94

으로써

0.93

POSITIVE LOGITS

SIDE

1.15

 וי

1.02

)){

1.02

পণ

0.96

𝘳

0.95

า

0.93

ুল

0.93

𝘭

0.92

𝘵

0.91

↵↵

0.91

Activations Density 0.055%