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

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

Neuron 1204 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: 0.4756. Min Range: -0.4756

Max Act: 0.4582. Min Act: -0.0000

Data Index: 1654760 (Python Code)

Max Activating Token Index: 634

Text #1

Max Range: 0.4756. Min Range: -0.4756

Max Act: 0.4756. Min Act: -0.0000

Data Index: 1370385 (Python Code)

Max Activating Token Index: 967

Text #2

Max Range: 0.4756. Min Range: -0.4756

Max Act: 0.3582. Min Act: -0.0000

Data Index: 743591 (C4 (Web Text))

Max Activating Token Index: 827

Text #3

Max Range: 0.4756. Min Range: -0.4756

Max Act: 0.3708. Min Act: -0.0000

Data Index: 1240243 (C4 (Web Text))

Max Activating Token Index: 313

Text #4

Max Range: 0.4756. Min Range: -0.4756

Max Act: 0.3901. Min Act: -0.0000

Data Index: 15671 (C4 (Web Text))

Max Activating Token Index: 356

Text #5

Max Range: 0.4756. Min Range: -0.4756

Max Act: 0.3460. Min Act: -0.0000

Data Index: 1190539 (C4 (Web Text))

Max Activating Token Index: 749

Text #6

Max Range: 0.4756. Min Range: -0.4756

Max Act: 0.3263. Min Act: -0.0000

Data Index: 1524599 (Python Code)

Max Activating Token Index: 137

Text #7

Max Range: 0.4756. Min Range: -0.4756

Max Act: 0.3751. Min Act: -0.0000

Data Index: 1325757 (C4 (Web Text))

Max Activating Token Index: 519

Text #8

Max Range: 0.4756. Min Range: -0.4756

Max Act: 0.3241. Min Act: -0.0000

Data Index: 97202 (C4 (Web Text))

Max Activating Token Index: 807

Text #9

Max Range: 0.4756. Min Range: -0.4756

Max Act: 0.3400. Min Act: -0.0000

Data Index: 541926 (C4 (Web Text))

Max Activating Token Index: 467

Text #10

Max Range: 0.4756. Min Range: -0.4756

Max Act: 0.4347. Min Act: -0.0000

Data Index: 1415514 (Python Code)

Max Activating Token Index: 658

Text #11

Max Range: 0.4756. Min Range: -0.4756

Max Act: 0.3375. Min Act: -0.0000

Data Index: 1382533 (Python Code)

Max Activating Token Index: 680

Text #12

Max Range: 0.4756. Min Range: -0.4756

Max Act: 0.3424. Min Act: -0.0000

Data Index: 1181469 (C4 (Web Text))

Max Activating Token Index: 364

Text #13

Max Range: 0.4756. Min Range: -0.4756

Max Act: 0.3175. Min Act: -0.0000

Data Index: 359421 (C4 (Web Text))

Max Activating Token Index: 707

Text #14

Max Range: 0.4756. Min Range: -0.4756

Max Act: 0.3312. Min Act: -0.0000

Data Index: 1571136 (Python Code)

Max Activating Token Index: 813

Text #15

Max Range: 0.4756. Min Range: -0.4756

Max Act: 0.3156. Min Act: -0.0000

Data Index: 889021 (C4 (Web Text))

Max Activating Token Index: 954

Text #16

Max Range: 0.4756. Min Range: -0.4756

Max Act: 0.3525. Min Act: -0.0000

Data Index: 141145 (C4 (Web Text))

Max Activating Token Index: 822

Text #17

Max Range: 0.4756. Min Range: -0.4756

Max Act: 0.3203. Min Act: -0.0000

Data Index: 1495322 (Python Code)

Max Activating Token Index: 323

Text #18

Max Range: 0.4756. Min Range: -0.4756

Max Act: 0.3229. Min Act: -0.0000

Data Index: 1381085 (Python Code)

Max Activating Token Index: 678

Text #19

Max Range: 0.4756. Min Range: -0.4756

Max Act: 0.3074. Min Act: -0.0000

Data Index: 1606249 (Python Code)

Max Activating Token Index: 432