INDEX

Explanations

pursuant to

The neuron detects occurrences of the legal bigram “pursuant to.”

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

are

-2.03

 will

-2.00

has

-1.89

щита

-1.80

 seluruh

-1.78

 didn

-1.77

 δύο

-1.77

使用的

-1.74

鳌

-1.73

])));

-1.72

POSITIVE LOGITS

マジで

2.23

 pérd

2.20

おめでとう

2.13

 土産

2.09

やっぱ

2.09

ホント

2.02

明日は

2.00

ほんと

1.98

鯔

1.91

viembre

1.90

Activations Density 0.004%