INDEX

Explanations

addition of

The neuron detects phrases of the form “the addition of …,” i.e. when text describes adding something.

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

}$

-1.67

姆

-1.65

呀

-1.58

昙

-1.56



-1.52

Lithuania

-1.52

":

-1.50

)$

-1.50

begin

-1.48

 остальные

-1.48

POSITIVE LOGITS

 deslig

2.28

2.17

 such

1.89

辚

1.84

 something

1.80

 perguntas

1.78

 semelh

1.77

 acidente

1.73

 ouvido

1.72

 vítimas

1.70

Activations Density 0.021%