INDEX

Explanations

Development of, Her law

The neuron fires strongly on closing‐delimiter tokens in code (e.g. ‘}’ or “);” that end statements or blocks).

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

 alkoh

-1.41

 Traducción

-1.15

 optik

-1.11

 klinik

-1.09

 mikrofon

-1.09

>*/

-1.07

 femininas

-1.07

 Drž

-1.06

 kafe

-1.05

ції

-1.02

POSITIVE LOGITS

1.24

and

1.19

on

1.15

1.05

}{

1.02

 ovviamente

0.99

0.95

0.94

0.93

 after

0.87

Activations Density 0.000%