INDEX

Explanations

ending punctuation

This neuron detects calls for user engagement or feedback (e.g. prompts to comment, ask for help, or contact).

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

 있는데

0.57

 repression

0.54

 없다

0.53

なのに

0.52

 ໜອງ

0.52

 দ্রব

0.52

 Therefore

0.51

 agonist

0.51

 suppresses

0.51

ğinden

0.51

POSITIVE LOGITS

:)

1.32

 🙂

1.32

😊

1.32

!</

1.28

 😊

1.23

🙂

1.22

1.20

:-)

1.18

☺️

1.17

😉

1.17

Activations Density 0.503%