INDEX

Explanations

questions with variables

The neuron fires on the commas that separate numeric list items (i.e. punctuation delimiting sequences of 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

Jangan

-2.27

Rumah

-2.25



-2.25

鉆

-2.25

䉴

-2.22

Mengapa

-2.20

Hanya

-2.16

Sekarang

-2.14

 برای

-2.11

Makanan

-2.11

POSITIVE LOGITS

快要

2.86

and

2.61

are

2.31

}$,

2.25

 cliché

2.14

)];

2.11



2.11

de

2.11

猝

2.03

tisp

2.02

Activations Density 0.127%