INDEX

Explanations

test followed by specific terms

The neuron strongly activates on occurrences of the token “Test” (in names of test functions, classes, or comments).

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

-1.45

-1.30

’

-1.27

igts

-1.27

 عنه

-1.25

</h2>

-1.23

-1.22

 ‘‘

-1.18

tă

-1.18

POSITIVE LOGITS

Nicht

1.57

of

1.53

 règne

1.52

 feltro

1.45

実際に

1.41

 réalisateur

1.40

今年は

1.39

 کرنے

1.37

 哪些

1.35

あなたが

1.34

Activations Density 0.038%