INDEX

Explanations

phrases like "x speaking"

The neuron detects multiword discourse‐marker phrases formed by an adverb or adjective plus “speaking” (e.g. “generally speaking,” “more broadly speaking”).

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

</b>

-1.74

 slew

-1.63

 使っ

-1.59

ască

-1.57

 editados

-1.55

 länge

-1.53

éroport

-1.52

ፕ

-1.51

她就

-1.48

BIÉN

-1.48

POSITIVE LOGITS

the

1.76

 gewähr

1.74

it

1.72

 ersche

1.72

there

1.69

«

1.68

“

1.66

if

1.64

you

1.63

Activations Density 0.009%