INDEX

Explanations

typo words

The neuron fires on malformed or misspelled word fragments.

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

もともと

-1.58

ano

-1.45

ia

-1.43

元々

-1.41

 बहुत

-1.38

Олег

-1.31

us

-1.30

ini

-1.28

！！！”

-1.27

-1.26

POSITIVE LOGITS

maxn

1.64

of

1.56

 tuvimos

1.41

bucks

1.40

 were

1.38

_;

1.36

preso

1.35

ContentAsync

1.35

teh

1.34

),

1.34

Activations Density 0.196%