INDEX

Explanations

usage, prerequisites, setup

This neuron fires on document section headings or titles (e.g. “Function:”, “Early life”, “Etymology”, “Overview”) rather than ordinary content.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

1.23

:}

1.19

：

1.15

:</

1.13

!:

1.08

:&

1.08

:";

1.02

?:

0.98

+:

0.97

$:

0.96

POSITIVE LOGITS

edit

1.01

 edit

0.95

Edit

0.93

 editar

0.86

 Edit

0.82

 править

0.81

editar

0.71

editing

0.71

编辑

0.70

",[

0.69

Activations Density 0.008%