Model: SoLU Model: 4 Layers, 2048 Neurons per Layer

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

Neuron 12 in Layer 1

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-4l')

Text #0

Max Range: 0.0347. Min Range: -0.0347

Max Act: 0.0340. Min Act: -0.0006

Data Index: 828969 (C4 (Web Text))

Max Activating Token Index: 208

Text #1

Max Range: 0.0347. Min Range: -0.0347

Max Act: 0.0317. Min Act: -0.0006

Data Index: 56957 (C4 (Web Text))

Max Activating Token Index: 12

Text #2

Max Range: 0.0347. Min Range: -0.0347

Max Act: 0.0347. Min Act: -0.0006

Data Index: 761388 (C4 (Web Text))

Max Activating Token Index: 708

Text #3

Max Range: 0.0347. Min Range: -0.0347

Max Act: 0.0333. Min Act: -0.0006

Data Index: 1017698 (C4 (Web Text))

Max Activating Token Index: 766

Text #4

Max Range: 0.0347. Min Range: -0.0347

Max Act: 0.0303. Min Act: -0.0006

Data Index: 591623 (C4 (Web Text))

Max Activating Token Index: 830

Text #5

Max Range: 0.0347. Min Range: -0.0347

Max Act: 0.0312. Min Act: -0.0006

Data Index: 143773 (C4 (Web Text))

Max Activating Token Index: 185

Text #6

Max Range: 0.0347. Min Range: -0.0347

Max Act: 0.0300. Min Act: -0.0006

Data Index: 1342128 (C4 (Web Text))

Max Activating Token Index: 870

Text #7

Max Range: 0.0347. Min Range: -0.0347

Max Act: 0.0306. Min Act: -0.0006

Data Index: 266431 (C4 (Web Text))

Max Activating Token Index: 876

Text #8

Max Range: 0.0347. Min Range: -0.0347

Max Act: 0.0312. Min Act: -0.0006

Data Index: 165845 (C4 (Web Text))

Max Activating Token Index: 184

Text #9

Max Range: 0.0347. Min Range: -0.0347

Max Act: 0.0304. Min Act: -0.0006

Data Index: 582455 (C4 (Web Text))

Max Activating Token Index: 796

Text #10

Max Range: 0.0347. Min Range: -0.0347

Max Act: 0.0322. Min Act: -0.0006

Data Index: 1448876 (Python Code)

Max Activating Token Index: 930

Text #11

Max Range: 0.0347. Min Range: -0.0347

Max Act: 0.0294. Min Act: -0.0006

Data Index: 878385 (C4 (Web Text))

Max Activating Token Index: 16

Text #12

Max Range: 0.0347. Min Range: -0.0347

Max Act: 0.0279. Min Act: -0.0006

Data Index: 1013645 (C4 (Web Text))

Max Activating Token Index: 770

Text #13

Max Range: 0.0347. Min Range: -0.0347

Max Act: 0.0279. Min Act: -0.0006

Data Index: 1304153 (C4 (Web Text))

Max Activating Token Index: 470

Text #14

Max Range: 0.0347. Min Range: -0.0347

Max Act: 0.0304. Min Act: -0.0006

Data Index: 1201203 (C4 (Web Text))

Max Activating Token Index: 850

Text #15

Max Range: 0.0347. Min Range: -0.0347

Max Act: 0.0296. Min Act: -0.0006

Data Index: 1320579 (C4 (Web Text))

Max Activating Token Index: 978

Text #16

Max Range: 0.0347. Min Range: -0.0347

Max Act: 0.0288. Min Act: -0.0006

Data Index: 1517013 (Python Code)

Max Activating Token Index: 501

Text #17

Max Range: 0.0347. Min Range: -0.0347

Max Act: 0.0279. Min Act: -0.0006

Data Index: 1174668 (C4 (Web Text))

Max Activating Token Index: 301

Text #18

Max Range: 0.0347. Min Range: -0.0347

Max Act: 0.0272. Min Act: -0.0006

Data Index: 1311618 (C4 (Web Text))

Max Activating Token Index: 726

Text #19

Max Range: 0.0347. Min Range: -0.0347

Max Act: 0.0278. Min Act: -0.0005

Data Index: 325573 (C4 (Web Text))

Max Activating Token Index: 649