INDEX

Explanations

Ernie

The neuron strongly activates on numeric tokens representing floating‐point values (decimal numbers).

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

たとえば

0.75

=-\

0.61

if

0.59

even

0.57

 mesela

0.57

 például

0.57

 misalnya

0.55

=\

0.55

tiny

0.55

wrong

0.54

POSITIVE LOGITS

 தற்போது

0.82

年来

0.73

 Collectively

0.72

いております

0.72

 сейчас

0.71

 입니다

0.71

 Coin

0.70

現在

0.69

 اکنون

0.68

mıştır

0.64

Activations Density 0.106%