INDEX

Explanations

command line execution showing output

This neuron activates on command‐line tokens (e.g. shell commands, flags, and pipeline components).

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

seys

-0.81

 popraw

-0.81

 crée

-0.79

vanta

-0.79

 muut

-0.78

taş

-0.77

γκ

-0.77

ρά

-0.76

oler

-0.75

ダル

-0.75

POSITIVE LOGITS

 output

1.42

showAll

1.34

 grep

1.25

输出

1.22

output

1.21

grep

1.11

Output

1.09

 showing

1.04

 Output

1.03

Activations Density 0.075%