INDEX

Explanations

source of

The neuron detects document‐metadata labels and citations—tokens like “Source” (and its following colon) that introduce references, resources, or other attribution information.

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

rj

0.64

 erscheinen

0.61

nake

0.61

submission

0.60

आज

0.59

 vorbe

0.59

each

0.57

 dissol

0.57

ustic

0.57

าที

0.57

POSITIVE LOGITS

 sources

1.61

源

1.56

 source

1.53

 Sources

1.53

 Source

1.52

 SOURCES

1.45

 SOURCE

1.42

 fuente

1.40

 nguồn

1.40

 źród

1.40

Activations Density 0.124%