INDEX

Explanations

customer service representatives

The main thing this neuron does is find mentions of customer‐service assistance (e.g. “customer care,” “representative,” “assist,” “assistance”).

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

for

-1.31

 also

-1.23

 falschen

-1.21

慢性

-1.15

 Septiembre

-1.15

 septembrie

-1.14

YI

-1.13

也

-1.13

 Temporada

-1.13

其他

-1.12

POSITIVE LOGITS

who

1.79

作る

1.44

理由は

1.37

 ήταν

1.32

ína

1.30

 piernas

1.28

èdent

1.27

ええ

1.27

1.25

Activations Density 0.037%