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

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

Neuron 4195 in Layer 6

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 2.7473. Min Range: -2.7473

Max Act: 2.7473. Min Act: -0.0000

Data Index: 233660 (C4 (Web Text))

Max Activating Token Index: 205

Text #1

Max Range: 2.7473. Min Range: -2.7473

Max Act: 1.7656. Min Act: -0.0000

Data Index: 896081 (C4 (Web Text))

Max Activating Token Index: 825

Text #2

Max Range: 2.7473. Min Range: -2.7473

Max Act: 1.7242. Min Act: -0.0000

Data Index: 388984 (C4 (Web Text))

Max Activating Token Index: 495

Text #3

Max Range: 2.7473. Min Range: -2.7473

Max Act: 1.7134. Min Act: -0.0000

Data Index: 62684 (C4 (Web Text))

Max Activating Token Index: 877

Text #4

Max Range: 2.7473. Min Range: -2.7473

Max Act: 1.6247. Min Act: -0.0000

Data Index: 983285 (C4 (Web Text))

Max Activating Token Index: 524

Text #5

Max Range: 2.7473. Min Range: -2.7473

Max Act: 1.5927. Min Act: -0.0000

Data Index: 156286 (C4 (Web Text))

Max Activating Token Index: 935

Text #6

Max Range: 2.7473. Min Range: -2.7473

Max Act: 1.4675. Min Act: -0.0000

Data Index: 1486322 (Python Code)

Max Activating Token Index: 509

Text #7

Max Range: 2.7473. Min Range: -2.7473

Max Act: 1.4704. Min Act: -0.0000

Data Index: 410293 (C4 (Web Text))

Max Activating Token Index: 1001

Text #8

Max Range: 2.7473. Min Range: -2.7473

Max Act: 1.2534. Min Act: -0.0000

Data Index: 1267720 (C4 (Web Text))

Max Activating Token Index: 345

Text #9

Max Range: 2.7473. Min Range: -2.7473

Max Act: 1.3039. Min Act: -0.0000

Data Index: 611115 (C4 (Web Text))

Max Activating Token Index: 1013

Text #10

Max Range: 2.7473. Min Range: -2.7473

Max Act: 1.3716. Min Act: -0.0000

Data Index: 1445898 (Python Code)

Max Activating Token Index: 941

Text #11

Max Range: 2.7473. Min Range: -2.7473

Max Act: 1.2529. Min Act: -0.0000

Data Index: 1307160 (C4 (Web Text))

Max Activating Token Index: 233

Text #12

Max Range: 2.7473. Min Range: -2.7473

Max Act: 1.3251. Min Act: -0.0000

Data Index: 596927 (C4 (Web Text))

Max Activating Token Index: 675

Text #13

Max Range: 2.7473. Min Range: -2.7473

Max Act: 1.1720. Min Act: -0.0000

Data Index: 355390 (C4 (Web Text))

Max Activating Token Index: 832

Text #14

Max Range: 2.7473. Min Range: -2.7473

Max Act: 1.0962. Min Act: -0.0000

Data Index: 1624763 (Python Code)

Max Activating Token Index: 981

Text #15

Max Range: 2.7473. Min Range: -2.7473

Max Act: 1.2345. Min Act: -0.0000

Data Index: 1457371 (Python Code)

Max Activating Token Index: 653

Text #16

Max Range: 2.7473. Min Range: -2.7473

Max Act: 1.3071. Min Act: -0.0000

Data Index: 1480839 (Python Code)

Max Activating Token Index: 331

Text #17

Max Range: 2.7473. Min Range: -2.7473

Max Act: 1.2364. Min Act: -0.0000

Data Index: 204924 (C4 (Web Text))

Max Activating Token Index: 794

Text #18

Max Range: 2.7473. Min Range: -2.7473

Max Act: 1.1543. Min Act: -0.0000

Data Index: 643887 (C4 (Web Text))

Max Activating Token Index: 475

Text #19

Max Range: 2.7473. Min Range: -2.7473

Max Act: 1.2309. Min Act: -0.0000

Data Index: 284590 (C4 (Web Text))

Max Activating Token Index: 643