INDEX

Explanations

specific word followed by another

The neuron fires most strongly on the more unusual, content‐bearing words (the longer or domain‐specific terms) in a sentence rather than common function words.

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

to

-1.63

𝐺

-1.37

eventually

-1.32

unno

-1.30

is

-1.30

给自己

-1.28

 brinquedos

-1.27

 there

-1.26

这般

-1.26

the

-1.25

POSITIVE LOGITS

Also

1.59

Our

1.49

 also

1.38

How

1.33

Where

1.33

的能力

1.24

 zawsze

1.23

 ARXIV

1.23

Other

1.23

 ovat

1.21

Activations Density 0.231%