Model: SoLU Model: 12 Layers, 6144 Neurons per Layer

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

Neuron 5800 in Layer 9

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-12l')

Text #0

Max Range: 0.2000. Min Range: -0.2000

Max Act: 0.2000. Min Act: -0.0000

Data Index: 486342 (C4 (Web Text))

Max Activating Token Index: 623

Text #1

Max Range: 0.2000. Min Range: -0.2000

Max Act: 0.1815. Min Act: -0.0000

Data Index: 430223 (C4 (Web Text))

Max Activating Token Index: 754

Text #2

Max Range: 0.2000. Min Range: -0.2000

Max Act: 0.1950. Min Act: -0.0000

Data Index: 17714 (C4 (Web Text))

Max Activating Token Index: 738

Text #3

Max Range: 0.2000. Min Range: -0.2000

Max Act: 0.1915. Min Act: -0.0000

Data Index: 832834 (C4 (Web Text))

Max Activating Token Index: 142

Text #4

Max Range: 0.2000. Min Range: -0.2000

Max Act: 0.1993. Min Act: -0.0000

Data Index: 1115489 (C4 (Web Text))

Max Activating Token Index: 609

Text #5

Max Range: 0.2000. Min Range: -0.2000

Max Act: 0.1742. Min Act: -0.0000

Data Index: 408672 (C4 (Web Text))

Max Activating Token Index: 726

Text #6

Max Range: 0.2000. Min Range: -0.2000

Max Act: 0.1751. Min Act: -0.0000

Data Index: 1043972 (C4 (Web Text))

Max Activating Token Index: 861

Text #7

Max Range: 0.2000. Min Range: -0.2000

Max Act: 0.1721. Min Act: -0.0000

Data Index: 1050909 (C4 (Web Text))

Max Activating Token Index: 75

Text #8

Max Range: 0.2000. Min Range: -0.2000

Max Act: 0.1721. Min Act: -0.0000

Data Index: 1003805 (C4 (Web Text))

Max Activating Token Index: 600

Text #9

Max Range: 0.2000. Min Range: -0.2000

Max Act: 0.1643. Min Act: -0.0000

Data Index: 1238845 (C4 (Web Text))

Max Activating Token Index: 95

Text #10

Max Range: 0.2000. Min Range: -0.2000

Max Act: 0.1702. Min Act: -0.0000

Data Index: 1145298 (C4 (Web Text))

Max Activating Token Index: 509

Text #11

Max Range: 0.2000. Min Range: -0.2000

Max Act: 0.1789. Min Act: -0.0000

Data Index: 295487 (C4 (Web Text))

Max Activating Token Index: 450

Text #12

Max Range: 0.2000. Min Range: -0.2000

Max Act: 0.1688. Min Act: -0.0000

Data Index: 1420581 (Python Code)

Max Activating Token Index: 1018

Text #13

Max Range: 0.2000. Min Range: -0.2000

Max Act: 0.1603. Min Act: -0.0000

Data Index: 621060 (C4 (Web Text))

Max Activating Token Index: 745

Text #14

Max Range: 0.2000. Min Range: -0.2000

Max Act: 0.1543. Min Act: -0.0000

Data Index: 652961 (C4 (Web Text))

Max Activating Token Index: 692

Text #15

Max Range: 0.2000. Min Range: -0.2000

Max Act: 0.1700. Min Act: -0.0000

Data Index: 275488 (C4 (Web Text))

Max Activating Token Index: 547

Text #16

Max Range: 0.2000. Min Range: -0.2000

Max Act: 0.1520. Min Act: -0.0000

Data Index: 657078 (C4 (Web Text))

Max Activating Token Index: 390

Text #17

Max Range: 0.2000. Min Range: -0.2000

Max Act: 0.1690. Min Act: -0.0000

Data Index: 795198 (C4 (Web Text))

Max Activating Token Index: 992

Text #18

Max Range: 0.2000. Min Range: -0.2000

Max Act: 0.1607. Min Act: -0.0000

Data Index: 903462 (C4 (Web Text))

Max Activating Token Index: 109

Text #19

Max Range: 0.2000. Min Range: -0.2000

Max Act: 0.1531. Min Act: -0.0000

Data Index: 453925 (C4 (Web Text))

Max Activating Token Index: 385