INDEX

Explanations

quantities and emphasis

The neuron mainly activates on numeric tokens—especially quantities, measurements, and prices.

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

 trochu

1.22

 কিছুটা

1.15

 biraz

1.15

 थोड़ा

1.15

 थोड़ा

1.14

 немного

1.11

 약간

1.06

 somewhat

1.05

 slightly

1.03

 Slightly

1.02

POSITIVE LOGITS

 dozens

1.78

 hundreds

1.76

 десятки

1.64

 огром

1.59

 thousands

1.54

数百

1.52

 শতাধিক

1.52

 enormes

1.52

thousands

1.52

 सैकड़ों

1.51

Activations Density 1.279%