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

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

Neuron 845 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: 2.6415. Min Range: -2.6415

Max Act: 2.5198. Min Act: -0.0000

Data Index: 920867 (C4 (Web Text))

Max Activating Token Index: 565

Text #1

Max Range: 2.6415. Min Range: -2.6415

Max Act: 2.6415. Min Act: -0.0000

Data Index: 108223 (C4 (Web Text))

Max Activating Token Index: 458

Text #2

Max Range: 2.6415. Min Range: -2.6415

Max Act: 2.5345. Min Act: -0.0000

Data Index: 410419 (C4 (Web Text))

Max Activating Token Index: 8

Text #3

Max Range: 2.6415. Min Range: -2.6415

Max Act: 2.4986. Min Act: -0.0000

Data Index: 1259803 (C4 (Web Text))

Max Activating Token Index: 438

Text #4

Max Range: 2.6415. Min Range: -2.6415

Max Act: 2.4888. Min Act: -0.0000

Data Index: 313668 (C4 (Web Text))

Max Activating Token Index: 56

Text #5

Max Range: 2.6415. Min Range: -2.6415

Max Act: 2.4617. Min Act: -0.0000

Data Index: 233390 (C4 (Web Text))

Max Activating Token Index: 332

Text #6

Max Range: 2.6415. Min Range: -2.6415

Max Act: 2.3438. Min Act: -0.0000

Data Index: 1041452 (C4 (Web Text))

Max Activating Token Index: 129

Text #7

Max Range: 2.6415. Min Range: -2.6415

Max Act: 2.3851. Min Act: -0.0000

Data Index: 32435 (C4 (Web Text))

Max Activating Token Index: 789

Text #8

Max Range: 2.6415. Min Range: -2.6415

Max Act: 2.1483. Min Act: -0.0000

Data Index: 1524041 (Python Code)

Max Activating Token Index: 239

Text #9

Max Range: 2.6415. Min Range: -2.6415

Max Act: 2.2986. Min Act: -0.0000

Data Index: 1112045 (C4 (Web Text))

Max Activating Token Index: 89

Text #10

Max Range: 2.6415. Min Range: -2.6415

Max Act: 2.3568. Min Act: -0.0000

Data Index: 840699 (C4 (Web Text))

Max Activating Token Index: 999

Text #11

Max Range: 2.6415. Min Range: -2.6415

Max Act: 2.1475. Min Act: -0.0000

Data Index: 1390743 (Python Code)

Max Activating Token Index: 551

Text #12

Max Range: 2.6415. Min Range: -2.6415

Max Act: 2.2080. Min Act: -0.0000

Data Index: 321605 (C4 (Web Text))

Max Activating Token Index: 520

Text #13

Max Range: 2.6415. Min Range: -2.6415

Max Act: 2.2372. Min Act: -0.0000

Data Index: 1226990 (C4 (Web Text))

Max Activating Token Index: 52

Text #14

Max Range: 2.6415. Min Range: -2.6415

Max Act: 2.1795. Min Act: -0.0000

Data Index: 1098629 (C4 (Web Text))

Max Activating Token Index: 484

Text #15

Max Range: 2.6415. Min Range: -2.6415

Max Act: 2.1492. Min Act: -0.0000

Data Index: 1379565 (Python Code)

Max Activating Token Index: 747

Text #16

Max Range: 2.6415. Min Range: -2.6415

Max Act: 2.0477. Min Act: -0.0000

Data Index: 1020384 (C4 (Web Text))

Max Activating Token Index: 35

Text #17

Max Range: 2.6415. Min Range: -2.6415

Max Act: 2.1453. Min Act: -0.0000

Data Index: 389272 (C4 (Web Text))

Max Activating Token Index: 18

Text #18

Max Range: 2.6415. Min Range: -2.6415

Max Act: 2.2811. Min Act: -0.0000

Data Index: 561871 (C4 (Web Text))

Max Activating Token Index: 860

Text #19

Max Range: 2.6415. Min Range: -2.6415

Max Act: 2.2024. Min Act: -0.0000

Data Index: 750355 (C4 (Web Text))

Max Activating Token Index: 279