INDEX

Explanations

numbers and asterisks

The neuron strongly activates on isolated digits used as reference or section/footnote markers.

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

 enzym

0.71

├──

0.68



0.68

 '-')

0.64

 tional

0.64

'-'

0.64

 compds

0.64

 NgModule

0.64

 solns

0.63

 调用

0.63

POSITIVE LOGITS

しかし

0.77

Political

0.73

更に

0.71

 Further

0.71

Further

0.71

However

0.68

 However

0.67

さらに

0.64

Financial

0.63

 Political

0.62

Activations Density 0.007%