INDEX

Explanations

numeric sequence

This neuron activates on standalone numeric tokens (individual digits or short numeral sequences).

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

 crowned

0.72

а

0.67

ס

0.64

0.63

 straightened

0.62

૬

0.62

 Spinach

0.61

 Pero

0.61

 Moulton

0.61

 STATES

0.60

POSITIVE LOGITS

️⃣

1.05

⃣

0.90

txt

0.72

st

0.69

Eo

0.68

Emp

0.65

Res

0.64

Bas

0.64

Eye

0.64

Activations Density 0.332%