INDEX

Explanations

words followed by specific continuations

The neuron fires on tokens that cue or immediately precede numeric data (e.g. commas, prepositions like “in” or “at,” quantifiers such as “most” or “more,” and similar markers around numbers).

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

が

-1.27

ที่

-1.25

を

-1.20

 like

-1.17

だったが

-1.13

 apartament

-1.09

に

-1.05

 leuk

-1.05

それを

-1.02

alnya

-1.02

POSITIVE LOGITS

 jeśli

1.37

 razones

1.37

inverno

1.31

ようで

1.30

ariConfig

1.24

感じで

1.23

unico

1.23

 TÉCN

1.22

icona

1.22

peteer

1.21

Activations Density 0.049%