INDEX

Explanations

question starters

This neuron fires on sentence boundaries and adjacent sentence-start words—e.g. it spikes on the period token and on sentence-initial tokens like “I,” “Is,” “Why,” etc.

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

Hulu

-0.95

ogener

-0.92

とその

-0.90

kennung

-0.87

 peru

-0.86

と同じ

-0.83

'>"

-0.82

 anno

-0.82

 jego

-0.81

 roulant

-0.81

POSITIVE LOGITS

 that

1.48

but

1.30

 this

1.23

確かに

1.19

 allegedly

1.13

so

1.10

 however

1.05

 وأن

1.05

 certainly

1.04

 Talvez

1.01

Activations Density 0.121%