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

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

Neuron 3988 in Layer 6

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 3.8667. Min Range: -3.8667

Max Act: 3.8667. Min Act: -0.0000

Data Index: 1468607 (Python Code)

Max Activating Token Index: 331

Text #1

Max Range: 3.8667. Min Range: -3.8667

Max Act: 3.5293. Min Act: -0.0000

Data Index: 1375287 (Python Code)

Max Activating Token Index: 42

Text #2

Max Range: 3.8667. Min Range: -3.8667

Max Act: 3.2223. Min Act: -0.0000

Data Index: 1633961 (Python Code)

Max Activating Token Index: 497

Text #3

Max Range: 3.8667. Min Range: -3.8667

Max Act: 3.2012. Min Act: -0.0000

Data Index: 1644111 (Python Code)

Max Activating Token Index: 762

Text #4

Max Range: 3.8667. Min Range: -3.8667

Max Act: 2.8709. Min Act: -0.0000

Data Index: 947924 (C4 (Web Text))

Max Activating Token Index: 804

Text #5

Max Range: 3.8667. Min Range: -3.8667

Max Act: 2.8405. Min Act: -0.0000

Data Index: 71023 (C4 (Web Text))

Max Activating Token Index: 817

Text #6

Max Range: 3.8667. Min Range: -3.8667

Max Act: 2.8421. Min Act: -0.0000

Data Index: 1379779 (Python Code)

Max Activating Token Index: 237

Text #7

Max Range: 3.8667. Min Range: -3.8667

Max Act: 2.7779. Min Act: -0.0000

Data Index: 202529 (C4 (Web Text))

Max Activating Token Index: 690

Text #8

Max Range: 3.8667. Min Range: -3.8667

Max Act: 2.8722. Min Act: -0.0000

Data Index: 807168 (C4 (Web Text))

Max Activating Token Index: 662

Text #9

Max Range: 3.8667. Min Range: -3.8667

Max Act: 2.6714. Min Act: -0.0000

Data Index: 940302 (C4 (Web Text))

Max Activating Token Index: 825

Text #10

Max Range: 3.8667. Min Range: -3.8667

Max Act: 2.6499. Min Act: -0.0000

Data Index: 628597 (C4 (Web Text))

Max Activating Token Index: 1015

Text #11

Max Range: 3.8667. Min Range: -3.8667

Max Act: 2.6637. Min Act: -0.0000

Data Index: 627374 (C4 (Web Text))

Max Activating Token Index: 156

Text #12

Max Range: 3.8667. Min Range: -3.8667

Max Act: 2.6518. Min Act: -0.0000

Data Index: 67607 (C4 (Web Text))

Max Activating Token Index: 10

Text #13

Max Range: 3.8667. Min Range: -3.8667

Max Act: 2.7883. Min Act: -0.0000

Data Index: 1539055 (Python Code)

Max Activating Token Index: 1012

Text #14

Max Range: 3.8667. Min Range: -3.8667

Max Act: 2.6298. Min Act: -0.0000

Data Index: 31376 (C4 (Web Text))

Max Activating Token Index: 9

Text #15

Max Range: 3.8667. Min Range: -3.8667

Max Act: 2.4981. Min Act: -0.0000

Data Index: 371963 (C4 (Web Text))

Max Activating Token Index: 382

Text #16

Max Range: 3.8667. Min Range: -3.8667

Max Act: 2.4911. Min Act: -0.0000

Data Index: 1208126 (C4 (Web Text))

Max Activating Token Index: 990

Text #17

Max Range: 3.8667. Min Range: -3.8667

Max Act: 2.4556. Min Act: -0.0000

Data Index: 828560 (C4 (Web Text))

Max Activating Token Index: 997

Text #18

Max Range: 3.8667. Min Range: -3.8667

Max Act: 2.5698. Min Act: -0.0000

Data Index: 180240 (C4 (Web Text))

Max Activating Token Index: 705

Text #19

Max Range: 3.8667. Min Range: -3.8667

Max Act: 2.5599. Min Act: -0.0000

Data Index: 900026 (C4 (Web Text))

Max Activating Token Index: 268