INDEX

Explanations

not or no

The neuron fires on contrastive or negation cues (e.g. “but,” “not,” “doesn’t,” “curiously”) that signal turning points or contradictions in text.

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

 тоже

-1.46

 också

-1.31

 również

-1.20

 навіть

-1.17

 głównie

-1.14

 nawet

-1.13

 jednod

-1.13

ناية

-1.11

 derfor

-1.10

 også

-1.08

POSITIVE LOGITS

 other

3.08

อื่น

2.25

其他

2.16

 остальных

2.13

other

2.02

 других

2.00

 другие

2.00

 others

1.96

 سایر

1.92

 остальные

1.91

Activations Density 0.102%