INDEX

Explanations

output

This neuron detects imperative, step‐by‐step instructions—particularly commands telling the reader to run, check, or interact with code.

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

 জায়গা

0.58

 desblo

0.56

名を

0.55

 suscrib

0.54

ോട്ടോ

0.54

 জায়গা

0.52

gestaltung

0.52

 общей

0.51

pository

0.51

 सास

0.50

POSITIVE LOGITS

 Output

1.60

输出

1.56

 输出

1.54

 output

1.52

 OUTPUT

1.52

 outputs

1.44

output

1.43

Output

1.41

 결과

1.41

OUTPUT

1.39

Activations Density 0.028%