INDEX

Explanations

introduces next item

The neuron detects discourse markers that introduce an additional point or example—most prominently the word “Another” (often in “Another way”).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 alberi

-1.92

with

-1.91

 ultimi

-1.78

-1.70

 lagos

-1.64

絰

-1.63

人

-1.62

不

-1.59

也

-1.56

 terakhir

-1.55

POSITIVE LOGITS

᠌

1.71

代わり

1.70

𖧵

1.66

 princípios

1.65

\_

1.63

鸶

1.62

 debes

1.60

 mooi

1.59

 thing

1.58

 corações

1.55

Activations Density 0.008%