Model: SoLU Model: 6 Layers, 3072 Neurons per Layer

Dataset: 80% C4 (Web Text) and 20% Python Code

Neuron 3041 in Layer 5

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-6l')

Text #0

Max Range: 3.8894. Min Range: -3.8894

Max Act: 3.8894. Min Act: -0.0001

Data Index: 839141 (C4 (Web Text))

Max Activating Token Index: 13

Text #1

Max Range: 3.8894. Min Range: -3.8894

Max Act: 0.9986. Min Act: -0.0000

Data Index: 999743 (C4 (Web Text))

Max Activating Token Index: 1020

Text #2

Max Range: 3.8894. Min Range: -3.8894

Max Act: 2.8173. Min Act: -0.0001

Data Index: 152905 (C4 (Web Text))

Max Activating Token Index: 83

Text #3

Max Range: 3.8894. Min Range: -3.8894

Max Act: 1.1150. Min Act: -0.0001

Data Index: 1108356 (C4 (Web Text))

Max Activating Token Index: 42

Text #4

Max Range: 3.8894. Min Range: -3.8894

Max Act: 0.8353. Min Act: -0.0001

Data Index: 1625950 (Python Code)

Max Activating Token Index: 796

Text #5

Max Range: 3.8894. Min Range: -3.8894

Max Act: 0.4609. Min Act: -0.0001

Data Index: 1446463 (Python Code)

Max Activating Token Index: 649

Text #6

Max Range: 3.8894. Min Range: -3.8894

Max Act: 0.3780. Min Act: -0.0001

Data Index: 1627829 (Python Code)

Max Activating Token Index: 489

Text #7

Max Range: 3.8894. Min Range: -3.8894

Max Act: 0.4304. Min Act: -0.0001

Data Index: 1089618 (C4 (Web Text))

Max Activating Token Index: 904

Text #8

Max Range: 3.8894. Min Range: -3.8894

Max Act: 0.5027. Min Act: -0.0001

Data Index: 1645200 (Python Code)

Max Activating Token Index: 600

Text #9

Max Range: 3.8894. Min Range: -3.8894

Max Act: 0.6711. Min Act: -0.0001

Data Index: 608153 (C4 (Web Text))

Max Activating Token Index: 1020

Text #10

Max Range: 3.8894. Min Range: -3.8894

Max Act: 0.3193. Min Act: -0.0001

Data Index: 1015012 (C4 (Web Text))

Max Activating Token Index: 1000

Text #11

Max Range: 3.8894. Min Range: -3.8894

Max Act: 0.1245. Min Act: -0.0001

Data Index: 1581578 (Python Code)

Max Activating Token Index: 642

Text #12

Max Range: 3.8894. Min Range: -3.8894

Max Act: 0.3960. Min Act: -0.0001

Data Index: 1555988 (Python Code)

Max Activating Token Index: 853

Text #13

Max Range: 3.8894. Min Range: -3.8894

Max Act: 0.0999. Min Act: -0.0001

Data Index: 1542065 (Python Code)

Max Activating Token Index: 617

Text #14

Max Range: 3.8894. Min Range: -3.8894

Max Act: 0.1802. Min Act: -0.0001

Data Index: 504320 (C4 (Web Text))

Max Activating Token Index: 946

Text #15

Max Range: 3.8894. Min Range: -3.8894

Max Act: 0.2518. Min Act: -0.0001

Data Index: 1530818 (Python Code)

Max Activating Token Index: 638

Text #16

Max Range: 3.8894. Min Range: -3.8894

Max Act: 0.2704. Min Act: -0.0001

Data Index: 1554334 (Python Code)

Max Activating Token Index: 1019

Text #17

Max Range: 3.8894. Min Range: -3.8894

Max Act: 0.6022. Min Act: -0.0001

Data Index: 1452901 (Python Code)

Max Activating Token Index: 709

Text #18

Max Range: 3.8894. Min Range: -3.8894

Max Act: 0.2737. Min Act: -0.0001

Data Index: 832133 (C4 (Web Text))

Max Activating Token Index: 996

Text #19

Max Range: 3.8894. Min Range: -3.8894

Max Act: 0.1728. Min Act: -0.0001

Data Index: 1531457 (Python Code)

Max Activating Token Index: 451