INDEX

Explanations

languages and punctuation

The neuron detects interrogative/question phrasing, firing strongly on question words (e.g. “how,” “what,” “is,” “to”) and question-mark contexts.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

čenje

-1.41

 vaše

-1.35

 zanim

-1.34

 svoju

-1.30

if

-1.29

 zápas

-1.27

 vrst

-1.27

 člove

-1.26

 objav

-1.25

 niekoľ

-1.23

POSITIVE LOGITS

„

1.31

»

1.29

 incluso

1.23

 Países

1.19

REM

1.19

 Unies

1.18

 REPRESENTATIVES

1.17

Š

1.16

 Aktivität

1.16

 regulador

1.16

Activations Density 0.073%