INDEX

Explanations

"i am" and "how to"

The neuron fires on the opening phrase of a user’s question—especially first‐person “I…” statements or leading interrogatives like “How…” at the start of a question.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 where

-1.60

 поэтому

-1.38

Fakta

-1.37

MENAFN

-1.31

Other

-1.31

Beskrivelse

-1.26

zess

-1.25

↵

-1.25

Ciclo

-1.24

 związku

-1.23

POSITIVE LOGITS

</strong>

1.48

 both

1.47

 jusqu

1.41

着一个

1.33

1.32

 chèvre

1.31

 jakarta

1.29

های

1.25

:]

1.24

ස

1.24

Activations Density 0.233%