INDEX

Explanations

Russian cyrillic endings

The neuron strongly activates on tokens containing foreign‐language or accented characters (i.e. non‐ASCII letters), signaling non-English words.

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

’

-2.70

 that

-2.19

 leggen

-2.17

 When

-2.05

 verschiedene

-2.00

 trekken

-1.89

 firent

-1.85

に来て

-1.83

…’

-1.80

 Мы

-1.77

POSITIVE LOGITS

และ

2.19

と

2.17

 jspb

2.14

ༀ

2.13

ly

2.11

ágenes

2.03

 животные

2.03

)$

2.00

↵↵

2.00

 neuer

1.99

Activations Density 0.045%