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

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

Neuron 3796 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: 0.4453. Min Range: -0.4453

Max Act: 0.4273. Min Act: 0.0000

Data Index: 1366896 (Python Code)

Max Activating Token Index: 398

Text #1

Max Range: 0.4453. Min Range: -0.4453

Max Act: 0.4453. Min Act: -0.0000

Data Index: 825235 (C4 (Web Text))

Max Activating Token Index: 266

Text #2

Max Range: 0.4453. Min Range: -0.4453

Max Act: 0.3587. Min Act: -0.0000

Data Index: 1654201 (Python Code)

Max Activating Token Index: 271

Text #3

Max Range: 0.4453. Min Range: -0.4453

Max Act: 0.4291. Min Act: -0.0000

Data Index: 537595 (C4 (Web Text))

Max Activating Token Index: 448

Text #4

Max Range: 0.4453. Min Range: -0.4453

Max Act: 0.4282. Min Act: 0.0000

Data Index: 1468429 (Python Code)

Max Activating Token Index: 805

Text #5

Max Range: 0.4453. Min Range: -0.4453

Max Act: 0.3277. Min Act: -0.0000

Data Index: 846196 (C4 (Web Text))

Max Activating Token Index: 266

Text #6

Max Range: 0.4453. Min Range: -0.4453

Max Act: 0.3483. Min Act: -0.0000

Data Index: 394653 (C4 (Web Text))

Max Activating Token Index: 797

Text #7

Max Range: 0.4453. Min Range: -0.4453

Max Act: 0.4242. Min Act: -0.0000

Data Index: 552722 (C4 (Web Text))

Max Activating Token Index: 677

Text #8

Max Range: 0.4453. Min Range: -0.4453

Max Act: 0.3952. Min Act: -0.0000

Data Index: 469147 (C4 (Web Text))

Max Activating Token Index: 122

Text #9

Max Range: 0.4453. Min Range: -0.4453

Max Act: 0.3685. Min Act: -0.0000

Data Index: 177795 (C4 (Web Text))

Max Activating Token Index: 180

Text #10

Max Range: 0.4453. Min Range: -0.4453

Max Act: 0.3461. Min Act: -0.0000

Data Index: 852282 (C4 (Web Text))

Max Activating Token Index: 829

Text #11

Max Range: 0.4453. Min Range: -0.4453

Max Act: 0.3294. Min Act: -0.0000

Data Index: 538738 (C4 (Web Text))

Max Activating Token Index: 283

Text #12

Max Range: 0.4453. Min Range: -0.4453

Max Act: 0.3355. Min Act: -0.0000

Data Index: 1162423 (C4 (Web Text))

Max Activating Token Index: 883

Text #13

Max Range: 0.4453. Min Range: -0.4453

Max Act: 0.3553. Min Act: -0.0000

Data Index: 1631081 (Python Code)

Max Activating Token Index: 856

Text #14

Max Range: 0.4453. Min Range: -0.4453

Max Act: 0.3599. Min Act: -0.0000

Data Index: 1555437 (Python Code)

Max Activating Token Index: 998

Text #15

Max Range: 0.4453. Min Range: -0.4453

Max Act: 0.3530. Min Act: -0.0000

Data Index: 582642 (C4 (Web Text))

Max Activating Token Index: 199

Text #16

Max Range: 0.4453. Min Range: -0.4453

Max Act: 0.3502. Min Act: -0.0000

Data Index: 434473 (C4 (Web Text))

Max Activating Token Index: 542

Text #17

Max Range: 0.4453. Min Range: -0.4453

Max Act: 0.3076. Min Act: -0.0000

Data Index: 1410118 (Python Code)

Max Activating Token Index: 960

Text #18

Max Range: 0.4453. Min Range: -0.4453

Max Act: 0.3437. Min Act: -0.0000

Data Index: 333467 (C4 (Web Text))

Max Activating Token Index: 831

Text #19

Max Range: 0.4453. Min Range: -0.4453

Max Act: 0.3367. Min Act: -0.0000

Data Index: 960138 (C4 (Web Text))

Max Activating Token Index: 398