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

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

Neuron 794 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.0389. Min Range: -0.0389

Max Act: 0.0389. Min Act: -0.0003

Data Index: 1028166 (C4 (Web Text))

Max Activating Token Index: 138

Text #1

Max Range: 0.0389. Min Range: -0.0389

Max Act: 0.0361. Min Act: -0.0003

Data Index: 1207238 (C4 (Web Text))

Max Activating Token Index: 621

Text #2

Max Range: 0.0389. Min Range: -0.0389

Max Act: 0.0340. Min Act: -0.0003

Data Index: 1082957 (C4 (Web Text))

Max Activating Token Index: 169

Text #3

Max Range: 0.0389. Min Range: -0.0389

Max Act: 0.0359. Min Act: -0.0003

Data Index: 324642 (C4 (Web Text))

Max Activating Token Index: 295

Text #4

Max Range: 0.0389. Min Range: -0.0389

Max Act: 0.0321. Min Act: -0.0002

Data Index: 68999 (C4 (Web Text))

Max Activating Token Index: 917

Text #5

Max Range: 0.0389. Min Range: -0.0389

Max Act: 0.0325. Min Act: -0.0003

Data Index: 445398 (C4 (Web Text))

Max Activating Token Index: 135

Text #6

Max Range: 0.0389. Min Range: -0.0389

Max Act: 0.0324. Min Act: -0.0003

Data Index: 631961 (C4 (Web Text))

Max Activating Token Index: 954

Text #7

Max Range: 0.0389. Min Range: -0.0389

Max Act: 0.0323. Min Act: -0.0004

Data Index: 356727 (C4 (Web Text))

Max Activating Token Index: 14

Text #8

Max Range: 0.0389. Min Range: -0.0389

Max Act: 0.0311. Min Act: -0.0003

Data Index: 268382 (C4 (Web Text))

Max Activating Token Index: 919

Text #9

Max Range: 0.0389. Min Range: -0.0389

Max Act: 0.0318. Min Act: -0.0002

Data Index: 74131 (C4 (Web Text))

Max Activating Token Index: 17

Text #10

Max Range: 0.0389. Min Range: -0.0389

Max Act: 0.0310. Min Act: -0.0003

Data Index: 1438162 (Python Code)

Max Activating Token Index: 792

Text #11

Max Range: 0.0389. Min Range: -0.0389

Max Act: 0.0325. Min Act: -0.0003

Data Index: 1138623 (C4 (Web Text))

Max Activating Token Index: 744

Text #12

Max Range: 0.0389. Min Range: -0.0389

Max Act: 0.0313. Min Act: -0.0003

Data Index: 1421166 (Python Code)

Max Activating Token Index: 546

Text #13

Max Range: 0.0389. Min Range: -0.0389

Max Act: 0.0309. Min Act: -0.0003

Data Index: 1441167 (Python Code)

Max Activating Token Index: 186

Text #14

Max Range: 0.0389. Min Range: -0.0389

Max Act: 0.0316. Min Act: -0.0003

Data Index: 750347 (C4 (Web Text))

Max Activating Token Index: 129

Text #15

Max Range: 0.0389. Min Range: -0.0389

Max Act: 0.0289. Min Act: -0.0003

Data Index: 1265896 (C4 (Web Text))

Max Activating Token Index: 779

Text #16

Max Range: 0.0389. Min Range: -0.0389

Max Act: 0.0308. Min Act: -0.0003

Data Index: 1429024 (Python Code)

Max Activating Token Index: 406

Text #17

Max Range: 0.0389. Min Range: -0.0389

Max Act: 0.0307. Min Act: -0.0002

Data Index: 857033 (C4 (Web Text))

Max Activating Token Index: 265

Text #18

Max Range: 0.0389. Min Range: -0.0389

Max Act: 0.0298. Min Act: -0.0004

Data Index: 593572 (C4 (Web Text))

Max Activating Token Index: 725

Text #19

Max Range: 0.0389. Min Range: -0.0389

Max Act: 0.0294. Min Act: -0.0003

Data Index: 486167 (C4 (Web Text))

Max Activating Token Index: 378