INDEX

Explanations

words followed by only

The neuron detects tokens that express conditions or qualifiers—words like “true,” “only,” “before,” “second,” and similar terms that set constraints or qualifications.

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

塒

-1.34

৳

-1.22

ridescent

-1.21

üksek

-1.20

 周年

-1.20

ண்டும்

-1.18

inerary

-1.14

吽

-1.13

akyti

-1.11

&=&

-1.06

POSITIVE LOGITS

 only

13.06

 только

9.25

only

8.50

Only

8.13

 tylko

7.78

 тільки

7.50

 فقط

7.50

 Only

7.47

 ONLY

7.09

 רק

6.75

Activations Density 1.612%