INDEX

Explanations

The neuron fires on version‐suffix tokens (e.g. “alpha,” “beta,” “rc,” etc.) in software version strings.

version numbers (alpha, beta, rc)

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

 from

-1.36

 that

-1.33

 without

-1.27

 after

-1.27

객체

-1.25

 during

-1.24

 prijzen

-1.24

compromising

-1.22

an

-1.21

 while

-1.20

POSITIVE LOGITS

1.71

1.45

Răsp

1.27

ឪ

1.18

 nimmer

1.14

 непосредственно

1.13

'',

1.13

iações

1.12

Instead

1.11

Activations Density 0.005%