INDEX

Explanations

who asking questions

The neuron fires on interrogative question prompts, especially sentences beginning with question words like “Who,” “Where,” etc.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 indist

0.71

 زائ

0.66

 irgendwie

0.61

 хотелось

0.61

 アップ

0.59

 vorhand

0.58

obium

0.58

 Pending

0.57

 istnie

0.57

 موجود

0.56

POSITIVE LOGITS

among

0.76

 among

0.74

 जानता

0.74

 owns

0.72

 dares

0.71

 knows

0.67

oley

0.67

ops

0.67

ceği

0.67

opee

0.66

Activations Density 0.124%