INDEX

Explanations

`and` followed by other words

This neuron doesn’t activate on any of the shown text—in other words, it appears to be effectively “dead” (it finds no patterns in these excerpts).

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

is

-1.95

are

-1.34

 that

-1.33

睥

-1.32

 sencilla

-1.32

 posando

-1.27

勿論

-1.24

の場合

-1.23

 there

-1.23

it

-1.23

POSITIVE LOGITS

 jemals

1.49

THING

1.48

$;

1.34

]--;

1.31

 всеми

1.31

EdgeInsets

1.27

 Tüm

1.26

 enraged

1.25

ibration

1.24

úgó

1.24

Activations Density 0.033%