INDEX

Explanations

specific nouns and concepts

The neuron strongly activates on single-word proper names—especially programming languages or database systems (e.g. Java, C#, SQL, MySQL) and sports terms (e.g. swimming, diving, water polo).

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

 each

-1.44

 sitten

-1.25

offie

-1.22

hva

-1.21

}/${

-1.19

 being

-1.19

Rumah

-1.18

ogn

-1.17

glichen

-1.14

Ness

-1.14

POSITIVE LOGITS

 когда

1.30

</td>

1.23

かつて

1.20

 kiedy

1.18

も

1.14

1.13

 üblichen

1.09

 ใน

1.09

Sep

1.09

 ਵਿੱਚ

1.08

Activations Density 0.301%