INDEX

Explanations

Returns:

This neuron detects caption‐style or metadata tokens common in social‐media or news snippets—e.g. “pic?”, “from Monday,” question marks, hashtags, and named‐entity tokens.

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

tabPadding

0.55

 chạm

0.52

 специально

0.51

 espes

0.50

幄

0.49

垛

0.49

يزات

0.49

abi

0.48

 specially

0.48

坛

0.48

POSITIVE LOGITS

pic

1.20

—

1.17

?—

1.14

.—

1.05

”—

1.05

pic

1.01

—

0.97

"—

0.97

(@

0.96

,—

0.96

Activations Density 0.053%