INDEX

Explanations

positive qualities and efficiency

The neuron activates on positive‐evaluation words that convey improvement, optimization, or desirable qualities (e.g. “healthy,” “efficient,” “ideal,” “quickly”).

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

<bos>

-1.09

每个

-0.91

ulauan

-0.89

 than

-0.87

婶

-0.86

 Некоторые

-0.85

 Эти

-0.84

 sometime

-0.82

 Такие

-0.82

egl

-0.80

POSITIVE LOGITS

 more

1.65

 יותר

1.52

 hơn

1.40

 zaidi

1.38

 daha

1.34

 davantage

1.33

 lebih

1.32

 più

1.22

更加

1.22

 més

1.18

Activations Density 0.067%