INDEX

Explanations

dates and measurements

The neuron never fires on any of these inputs—it appears to be effectively “dead” and does not detect any pattern.

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

～～～

-1.47

！(

-1.29

 especiais

-1.19

出来ない

-1.18

 kvad

-1.17

now

-1.17

いません

-1.16

 klara

-1.16

ázaro

-1.16

<bos>

-1.15

POSITIVE LOGITS

際に

1.28

1.20

1.18

zą

1.16

為に

1.15

 podendo

1.15

1.14

rof

1.13

就已经

1.10

ena

1.07

Activations Density 0.000%