INDEX

Explanations

have to _ verb

The neuron fires on content (action) verbs in their base form (e.g. “notice,” “say,” “pay,” “wait,” “go,” “cope,” “start,” “guess,” “solve”).

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

 unangemessene

-1.38

 arbeitet

-1.30

什么事

-1.21

告诉你

-1.20

 need

-1.19

 corporativa

-1.14

怎么可能

-1.14

第一时间

-1.13

違って

-1.13

 Auffassung

-1.12

POSITIVE LOGITS

 some

1.52

to

1.36

and

1.23

 sitten

1.19

 elämä

1.18

 every

1.17

 these

1.16

 their

1.12

 Haushalt

1.10

 sezonu

1.10

Activations Density 0.066%