Model: SoLU Model: 10 Layers, 5120 Neurons per Layer

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

Neuron 1635 in Layer 9

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-10l')

Text #0

Max Range: 2.4328. Min Range: -2.4328

Max Act: 2.4328. Min Act: -0.0000

Data Index: 1550670 (Python Code)

Max Activating Token Index: 204

Text #1

Max Range: 2.4328. Min Range: -2.4328

Max Act: 1.9372. Min Act: -0.0000

Data Index: 1395302 (Python Code)

Max Activating Token Index: 657

Text #2

Max Range: 2.4328. Min Range: -2.4328

Max Act: 2.0679. Min Act: -0.0000

Data Index: 482137 (C4 (Web Text))

Max Activating Token Index: 948

Text #3

Max Range: 2.4328. Min Range: -2.4328

Max Act: 1.9086. Min Act: -0.0000

Data Index: 1591279 (Python Code)

Max Activating Token Index: 354

Text #4

Max Range: 2.4328. Min Range: -2.4328

Max Act: 2.0197. Min Act: -0.0000

Data Index: 1062163 (C4 (Web Text))

Max Activating Token Index: 702

Text #5

Max Range: 2.4328. Min Range: -2.4328

Max Act: 2.0436. Min Act: -0.0000

Data Index: 704533 (C4 (Web Text))

Max Activating Token Index: 967

Text #6

Max Range: 2.4328. Min Range: -2.4328

Max Act: 1.9329. Min Act: -0.0000

Data Index: 631591 (C4 (Web Text))

Max Activating Token Index: 121

Text #7

Max Range: 2.4328. Min Range: -2.4328

Max Act: 1.9960. Min Act: -0.0000

Data Index: 1276114 (C4 (Web Text))

Max Activating Token Index: 806

Text #8

Max Range: 2.4328. Min Range: -2.4328

Max Act: 1.7063. Min Act: -0.0000

Data Index: 50105 (C4 (Web Text))

Max Activating Token Index: 569

Text #9

Max Range: 2.4328. Min Range: -2.4328

Max Act: 1.7168. Min Act: -0.0000

Data Index: 597972 (C4 (Web Text))

Max Activating Token Index: 455

Text #10

Max Range: 2.4328. Min Range: -2.4328

Max Act: 1.5758. Min Act: -0.0000

Data Index: 684781 (C4 (Web Text))

Max Activating Token Index: 80

Text #11

Max Range: 2.4328. Min Range: -2.4328

Max Act: 1.8707. Min Act: -0.0000

Data Index: 1494670 (Python Code)

Max Activating Token Index: 35

Text #12

Max Range: 2.4328. Min Range: -2.4328

Max Act: 1.6639. Min Act: -0.0000

Data Index: 614707 (C4 (Web Text))

Max Activating Token Index: 632

Text #13

Max Range: 2.4328. Min Range: -2.4328

Max Act: 1.6170. Min Act: -0.0000

Data Index: 1618910 (Python Code)

Max Activating Token Index: 203

Text #14

Max Range: 2.4328. Min Range: -2.4328

Max Act: 1.6121. Min Act: -0.0000

Data Index: 1349838 (C4 (Web Text))

Max Activating Token Index: 200

Text #15

Max Range: 2.4328. Min Range: -2.4328

Max Act: 1.5895. Min Act: -0.0000

Data Index: 1024383 (C4 (Web Text))

Max Activating Token Index: 620

Text #16

Max Range: 2.4328. Min Range: -2.4328

Max Act: 1.6647. Min Act: -0.0000

Data Index: 1365555 (Python Code)

Max Activating Token Index: 40

Text #17

Max Range: 2.4328. Min Range: -2.4328

Max Act: 1.6698. Min Act: -0.0000

Data Index: 386395 (C4 (Web Text))

Max Activating Token Index: 314

Text #18

Max Range: 2.4328. Min Range: -2.4328

Max Act: 1.3742. Min Act: -0.0000

Data Index: 1625148 (Python Code)

Max Activating Token Index: 92

Text #19

Max Range: 2.4328. Min Range: -2.4328

Max Act: 1.5330. Min Act: -0.0000

Data Index: 1576639 (Python Code)

Max Activating Token Index: 316