INDEX

Explanations

most followed by intensifier

The neuron selectively detects the word “most,” i.e. the superlative degree marker.

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

and

-1.70

<bos>

-1.51

咐

-1.50

but

-1.49

it

-1.45

ándor

-1.42

 since

-1.31

はその

-1.27

けどね

-1.27

one

-1.26

POSITIVE LOGITS

还有什么

1.63

 importantly

1.63

’

1.59

是什么

1.55

termilk

1.50

the

1.45

もう一度

1.40

时候

1.40

当然是

1.39

Välkommen

1.38

Activations Density 0.021%