INDEX

Explanations

bolded list items

The neuron is detecting numeric tokens, especially numbers with decimal points.

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

 這個

0.71

uur

0.68

 ইলেক্ট্রোড

0.67

 조직

0.62

 Wheels

0.60

 Developments

0.60

UG

0.59

 Variety

0.59

 Research

0.58

 этом

0.58

POSITIVE LOGITS

 empathetic

0.64

 testAvg

0.57

 compassion

0.55

 maaf

0.55

 tenia

0.54

了一眼

0.54

 considerate

0.52

 njen

0.52

 waktu

0.52

cif

0.52

Activations Density 0.031%