INDEX

Explanations

court citations and names

The neuron fires on the numeric line‐range indicators in unified diffs (the “–start,count” and “+start,count” hunk headers).

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

 also

-1.67

-1.59

as

-1.58

to

-1.49

 only

-1.35

 three

-1.34

 Allgemeinen

-1.34

ための

-1.32

 instead

-1.32

틈

-1.30

POSITIVE LOGITS

konstruktion

1.25

ghter

1.25

 konkur

1.19

 beiden

1.16

 waarom

1.16

 JUSTICE

1.15

lorin

1.15

婳

1.14

tanleria

1.14

writeField

1.13

Activations Density 0.059%