INDEX

Explanations

comparisons using than

The neuron strongly activates on numeric tokens (especially three‐digit 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

of

-1.16

 also

-1.05

愎

-1.02

歌唱

-1.01

 group

-0.97

 development

-0.96

 other

-0.95

思想

-0.93

 several

-0.93

POSITIVE LOGITS

 this

1.71

 what

1.21

久しぶりの

1.13

妩

1.12

昨日は

1.10

 вдруг

1.09

久しぶりに

1.08

この日

1.07

昨日の

1.06

柔软

1.06

Activations Density 0.024%