INDEX

Explanations

good and not evil

The neuron fires on occurrences of the adjective “good” and its immediate arguments (e.g. “good will,” “good science,” “good Taliban,” “good case,” “good kings,” “good…and not evil,” 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

НР

-1.21

 sowohl

-1.20

 verwe

-1.14

движи

-1.13

 női

-1.11

𝘦

-1.09

ЕН

-1.09

𝘩

-1.05

XYGEN

-1.05

ід

-1.05

POSITIVE LOGITS

good

1.57

 good

1.52

Good

1.41

 still

1.27

 Good

1.24

 razões

1.21

↵↵

1.20

 verwendeten

1.15

great

1.15

 sés

1.14

Activations Density 0.091%