INDEX

Explanations

logical connectors and explanations

The neuron detects paragraph- or sentence-level transitional adverbial cues (e.g. “Recently,” “Additionally,” “First,” “Specifically,” etc.) that signal shifts or new points.

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

 they

-1.89

it

-1.69

遑

-1.50

erunner

-1.39

之以

-1.35

それが

-1.34

🏡

-1.30

 그것

-1.30

ﬀ

-1.27

 soms

-1.27

POSITIVE LOGITS

 Since

1.69

 although

1.62

 despite

1.52

 Following

1.41

 Although

1.38

Although

1.34

 However

1.34

 Despite

1.30

 since

1.29

 Interestingly

1.23

Activations Density 0.037%