INDEX

Explanations

constructor and destructor

The neuron activates on the “ctor” token—i.e. it picks out words containing the “ctor” suffix (as in destructor, attractor, etc.).

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

了一个

-3.09



-2.75

wortet

-2.53

潯

-2.53

ᯅ

-2.42

葑

-2.38

ology

-2.33

 insectes

-2.28

 修理

-2.22

真的很

-2.20

POSITIVE LOGITS

”;

2.86

Ꮏ

2.70

medriver

2.58

</h3>

2.53

↵

2.38

“

2.20

即便

2.19

</blockquote>

2.13

 determinar

2.09

 unearth

2.06

Activations Density 0.020%