INDEX

Explanations

compound words and phrases

The neuron reliably lights up on the “stems” of compound identifiers or multi‐part names (e.g. the first segment of hyphenated chemical names, filesystem calls like “open,” language‐code fragments like “latn,” album/taxon stems, etc.).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

kopi

-1.22

niet

-1.22

 flesta

-1.17

 favoris

-1.16

Mô

-1.15

 neutre

-1.13

₽

-1.13

moet

-1.12

 tradis

-1.12

 miljoner

-1.11

POSITIVE LOGITS

↵↵↵↵

1.46

↵↵↵↵↵↵↵↵↵↵

1.41

↵↵↵↵↵↵↵↵↵↵↵

1.40

1.39

1.38

↵↵↵↵↵↵↵↵↵↵↵↵↵↵

1.37

1.34

↵↵↵↵↵↵↵↵↵↵↵↵↵

1.30

1.29

Activations Density 0.084%