INDEX

Explanations

Eureka, Gamma, Share, Adventures

The neuron strongly activates on the “.” tokens used in code qualified names (i.e. the member‐access or namespace separators in programming code).

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

 with

-1.60

by

-1.58

 each

-1.43

to

-1.42

 every

-1.38

at

-1.34

but

-1.34

家は

-1.34

all

-1.33

 their

-1.32

POSITIVE LOGITS

 trám

1.69

感じです

1.38

で見る

1.36

 ovviamente

1.35

ようです

1.35

 regolare

1.32

様な

1.31

為に

1.30

んですよ

1.30

 bát

1.29

Activations Density 0.000%