INDEX

Explanations

consequences of

The neuron activates on occurrences of the word “consequence” (especially its plural “consequences”), flagging mentions of outcomes or effects.

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

 This

-3.03

</i>

-2.77

 Перейти

-2.70

所有人都

-2.53

</h1>

-2.38

 asesin

-2.33

derdag

-2.30

";

-2.25

身材

-2.25

?"

-2.19

POSITIVE LOGITS

</em>

2.84

</h2>

2.83

！！”

2.53

！！！”

2.48

～。

2.25

↵

2.25

чие

2.16

 Sénégal

2.11

!");

2.09

呖

2.09

Activations Density 0.009%