large numbers followed by time

The neuron detects large-scale numeric quantifiers (e.g. “thousand,” “million,” “hundred”).

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

-3.06

-2.75

-2.64

-2.61

-2.59

-2.50

赑

-2.33

-2.31

-2.19

-2.13

POSITIVE LOGITS

遄

2.16

Ꮨ

2.13

 reaf

2.08

鑪

2.05

iocene

2.03

ghalaya

2.02

 profesional

1.96

ynchrotron

1.95

InstrumentedTest

1.95

ga

1.94

Activations Density 0.007%