INDEX

Explanations

Q: question or The followed by utility/object/inference

This neuron activates on C++ (and similar) namespace or class qualifiers using the scope‐resolution operator “::.”

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 then

-1.32

-1.17

();

-1.13

ホイール

-1.07

 just

-1.02

 everyone

-1.02

然后

-1.02

чный

-1.01

no

-1.00

一隻

-0.99

POSITIVE LOGITS

 Wartawan

1.33

Wady

1.20

 thép

1.17

われている

1.16

ihnachten

1.12

⅝

1.12

 CORRECT

1.11

TIA

1.11

⅜

1.11

枚目

1.09

Activations Density 0.001%