INDEX

Explanations

questions and answers

This neuron detects interrogative content—especially question words (“who,” “what,” etc.) and the question‐mark punctuation signaling that a sentence is a question.

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

 appunto

0.64

 নানান

0.62

 当然

0.62

 thôi

0.61

ہی

0.61

 напротив

0.61

 각종

0.60

 тях

0.59

—

0.59

 алыңыз

0.57

POSITIVE LOGITS

 __________

1.27

 ________

1.26

 ____________

1.23

 _____________

1.22

 _______

1.20

__________

1.19

 ______

1.19

________

1.18

 ____

1.18

_______

1.18

Activations Density 0.084%