INDEX

Explanations

latex document classes and packages

The neuron selectively activates on numeric tokens—especially long numbers, identifiers, or decimal values.

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

Это

0.76

е

0.74

0.73

Flatten

0.71

 dint

0.71

 Nullam

0.70

 abut

0.68

Nic

0.67

Found

0.67

POSITIVE LOGITS

 розы

0.81

𝘪

0.80

 imigr

0.79

 alquiler

0.78

 pokuš

0.78

 católica

0.76

 estamp

0.75

 हराकर

0.75

 pasajeros

0.75

𝘰

0.75

Activations Density 0.003%