INDEX

Explanations

The neuron fires strongly at the document- or sentence-start token (i.e. it detects the beginning of an input sequence).

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

কিছু

0.28

 όλα

0.27

ല്ലാതെ

0.27

 unbelievable

0.26

 Gefühl

0.25

siehe

0.25

 यादगार

0.24

 ذریع

0.24

ूठी

0.24

 Faça

0.24

POSITIVE LOGITS

and

0.61

и

0.60

và

0.57

 или

0.53

 ಮತ್ತು

0.52

 மற்றும்

0.52

 και

0.51

ও

0.50

و

0.50

or

0.49

Activations Density 18.428%