INDEX

Explanations

a followed by a noun

The neuron fires on standalone numeric modifiers (cardinal numbers) used to quantify or specify things (e.g. “A 232-second season,” “an 81-point advisory,” “a 200-memorial service,” “a 96-gala dinner,” etc.).

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

っていない

-1.42

 garantiza

-1.37

が違う

-1.33

 estrena

-1.30

が無い

-1.25

 Tipos

-1.25

 incluye

-1.21

 capability

-1.21

ってない

-1.20

 possibility

-1.18

POSITIVE LOGITS

 will

1.69

and

1.30

at

1.16

：

1.12

 باشیم

1.11

🍁

1.10

🌸

1.09

 bermanfaat

1.05

ай

1.05

يور

1.04

Activations Density 0.001%