INDEX

Explanations

The neuron fires on tokens that begin a new sentence or segment—i.e. it picks out sentence‐initial words.

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

 जातात

1.29

becomes

1.28

 becomes

1.27

 می‌شوند

1.27

 vengono

1.23

しまいます

1.20

 begint

1.20

 становится

1.19

 нравится

1.19

 становятся

1.19

POSITIVE LOGITS

 buvo

2.70

 ήταν

2.64

 وكانت

2.59

 הייתה

2.59

 была

2.56

 była

2.56

 était

2.55

did

2.50

 היה

2.48

 كانت

2.47

Activations Density 1.767%