INDEX

Explanations

Quincy, ngx, clr, Ringo, Phase

The neuron fires on capitalized names and other proper‐noun tokens (e.g. Quincy, Ringo, module names like “ngx” or namespace prefixes like “clr”)—in other words, it identifies tokens that are proper nouns or acronyms.

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

 weißem

-2.81

不该

-2.66

 エントランス

-2.58

 velké

-2.53

りました

-2.45

我还是

-2.38

breviations

-2.38

我没

-2.31

 verfügb

-2.31

 الفلسطيني

-2.28

POSITIVE LOGITS

’

3.58

↵↵

2.75

↵

2.72

ен

2.39

2.27

</h2>

2.25

2.20

一个

2.06

 vous

2.06

«

2.06

Activations Density 0.003%