INDEX

Explanations

pronouns and possessives

The neuron strongly activates on personal names (and related subject-role words like “Mr,” “was,” “has,” “being,” “his,” etc.)—in other words, it’s picking out mentions of people (proper names and their pronouns/verbs) serving as sentence subjects.

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

 mukaan

-1.49

 however

-1.44

</b>

-1.37

﹔

-1.33

 oferec

-1.32

 フリル

-1.32

-1.31

of

-1.27

 cérami

-1.21

 daer

-1.20

POSITIVE LOGITS

 který

1.44

})));

1.38

 votre

1.36

two

1.32

had

1.32

 four

1.28

 Internasional

1.27

 hergestellt

1.26

 который

1.22

 другими

1.21

Activations Density 0.106%