INDEX

Explanations

superscripted numbers or citations

The neuron detects inline reference or footnote markers (e.g. bracketed superscript citations and numbered footnote indicators).

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

to

-1.89

and

-1.20

 from

-1.19

に使

-1.16

too

-1.15

-1.13

for

-1.12

-1.09

-1.06

 também

-1.03

POSITIVE LOGITS

<bos>

1.31

givers

1.18

pués

1.18

spectos

1.11

timme

1.10

そして

1.07

 felpa

1.06

auré

1.06

 manquer

1.06

1.05

Activations Density 0.009%