INDEX

Explanations

optional followed by punctuation

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 waarbij

1.84

함으로써

1.82

사를

1.71

 สาว

1.71

 αφού

1.70

 buscando

1.67

 quienes

1.65

 envolvendo

1.59

 olup

1.58

 Yayın

1.57

POSITIVE LOGITS

 stabilise

1.40

but

1.32

但不

1.22

也要

1.16

 stabilisation

1.14

 behaviour

1.14

 anaemia

1.13

 localised

1.11

 neighbourhood

1.10

 idea

1.10

Activations Density 0.103%