Model: SoLU Model: 3 Layers, 2048 Neurons per Layer

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

Neuron 253 in Layer 1

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-3l')

Text #0

Max Range: 0.0245. Min Range: -0.0245

Max Act: 0.0242. Min Act: -0.0006

Data Index: 634972 (C4 (Web Text))

Max Activating Token Index: 200

Text #1

Max Range: 0.0245. Min Range: -0.0245

Max Act: 0.0245. Min Act: -0.0006

Data Index: 1480661 (Python Code)

Max Activating Token Index: 972

Text #2

Max Range: 0.0245. Min Range: -0.0245

Max Act: 0.0233. Min Act: -0.0006

Data Index: 978272 (C4 (Web Text))

Max Activating Token Index: 193

Text #3

Max Range: 0.0245. Min Range: -0.0245

Max Act: 0.0221. Min Act: -0.0006

Data Index: 442854 (C4 (Web Text))

Max Activating Token Index: 781

Text #4

Max Range: 0.0245. Min Range: -0.0245

Max Act: 0.0211. Min Act: -0.0006

Data Index: 1175103 (C4 (Web Text))

Max Activating Token Index: 761

Text #5

Max Range: 0.0245. Min Range: -0.0245

Max Act: 0.0215. Min Act: -0.0006

Data Index: 856906 (C4 (Web Text))

Max Activating Token Index: 795

Text #6

Max Range: 0.0245. Min Range: -0.0245

Max Act: 0.0209. Min Act: -0.0006

Data Index: 897128 (C4 (Web Text))

Max Activating Token Index: 81

Text #7

Max Range: 0.0245. Min Range: -0.0245

Max Act: 0.0214. Min Act: -0.0006

Data Index: 1527512 (Python Code)

Max Activating Token Index: 897

Text #8

Max Range: 0.0245. Min Range: -0.0245

Max Act: 0.0223. Min Act: -0.0006

Data Index: 339166 (C4 (Web Text))

Max Activating Token Index: 557

Text #9

Max Range: 0.0245. Min Range: -0.0245

Max Act: 0.0215. Min Act: -0.0006

Data Index: 442449 (C4 (Web Text))

Max Activating Token Index: 760

Text #10

Max Range: 0.0245. Min Range: -0.0245

Max Act: 0.0215. Min Act: -0.0006

Data Index: 128069 (C4 (Web Text))

Max Activating Token Index: 197

Text #11

Max Range: 0.0245. Min Range: -0.0245

Max Act: 0.0214. Min Act: -0.0006

Data Index: 1151715 (C4 (Web Text))

Max Activating Token Index: 697

Text #12

Max Range: 0.0245. Min Range: -0.0245

Max Act: 0.0209. Min Act: -0.0006

Data Index: 1424866 (Python Code)

Max Activating Token Index: 320

Text #13

Max Range: 0.0245. Min Range: -0.0245

Max Act: 0.0204. Min Act: -0.0006

Data Index: 1015979 (C4 (Web Text))

Max Activating Token Index: 423

Text #14

Max Range: 0.0245. Min Range: -0.0245

Max Act: 0.0217. Min Act: -0.0006

Data Index: 1615604 (Python Code)

Max Activating Token Index: 406

Text #15

Max Range: 0.0245. Min Range: -0.0245

Max Act: 0.0203. Min Act: -0.0006

Data Index: 329830 (C4 (Web Text))

Max Activating Token Index: 756

Text #16

Max Range: 0.0245. Min Range: -0.0245

Max Act: 0.0198. Min Act: -0.0006

Data Index: 828513 (C4 (Web Text))

Max Activating Token Index: 784

Text #17

Max Range: 0.0245. Min Range: -0.0245

Max Act: 0.0198. Min Act: -0.0006

Data Index: 658148 (C4 (Web Text))

Max Activating Token Index: 646

Text #18

Max Range: 0.0245. Min Range: -0.0245

Max Act: 0.0212. Min Act: -0.0006

Data Index: 1571516 (Python Code)

Max Activating Token Index: 65

Text #19

Max Range: 0.0245. Min Range: -0.0245

Max Act: 0.0215. Min Act: -0.0006

Data Index: 1484851 (Python Code)

Max Activating Token Index: 649