INDEX

Explanations

counting numbers

the neuron is activated by standalone numeric tokens (numbers).

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

 Gründen

-1.69

 fixé

-1.66

９

-1.62

 remplacé

-1.61

 Когда

-1.60

 précisé

-1.59

 fianco

-1.58

 veicolo

-1.58

 Также

-1.57

 expliqué

-1.57

POSITIVE LOGITS

1.96

“

1.88

 kwamba

1.66

𖤣

1.62

in

1.60

只想

1.59

何より

1.59

 with

1.54

out

1.54

 from

1.52

Activations Density 0.136%