INDEX

Explanations

jumbled, scrambled, fragmented

The neuron detects terms referring to mixing up or breaking apart and then reassembling parts (e.g. scrambled, jumbled, fragmented, broken, pieces, unscrambled).

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

 desired

-1.51

by

-1.38

 need

-1.37

 same

-1.37

has

-1.30

 ghiaccio

-1.30

<i>

-1.27

 ialah

-1.24

↵↵

-1.23

 utilizzo

-1.23

POSITIVE LOGITS

笮

1.83

 můžete

1.64

 comien

1.38

茘

1.35

 således

1.34

 máte

1.34

፝

1.33

常见的

1.31

 mélange

1.30

 malheure

1.30

Activations Density 0.051%