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

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

Neuron 2018 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-10l')

Text #0

Max Range: 0.0507. Min Range: -0.0507

Max Act: 0.0507. Min Act: -0.0003

Data Index: 85234 (C4 (Web Text))

Max Activating Token Index: 888

Text #1

Max Range: 0.0507. Min Range: -0.0507

Max Act: 0.0468. Min Act: -0.0002

Data Index: 685111 (C4 (Web Text))

Max Activating Token Index: 465

Text #2

Max Range: 0.0507. Min Range: -0.0507

Max Act: 0.0429. Min Act: -0.0002

Data Index: 376329 (C4 (Web Text))

Max Activating Token Index: 68

Text #3

Max Range: 0.0507. Min Range: -0.0507

Max Act: 0.0364. Min Act: -0.0002

Data Index: 752609 (C4 (Web Text))

Max Activating Token Index: 205

Text #4

Max Range: 0.0507. Min Range: -0.0507

Max Act: 0.0398. Min Act: -0.0002

Data Index: 624398 (C4 (Web Text))

Max Activating Token Index: 420

Text #5

Max Range: 0.0507. Min Range: -0.0507

Max Act: 0.0353. Min Act: -0.0002

Data Index: 737913 (C4 (Web Text))

Max Activating Token Index: 421

Text #6

Max Range: 0.0507. Min Range: -0.0507

Max Act: 0.0315. Min Act: -0.0002

Data Index: 1286656 (C4 (Web Text))

Max Activating Token Index: 782

Text #7

Max Range: 0.0507. Min Range: -0.0507

Max Act: 0.0308. Min Act: -0.0002

Data Index: 1358115 (C4 (Web Text))

Max Activating Token Index: 212

Text #8

Max Range: 0.0507. Min Range: -0.0507

Max Act: 0.0321. Min Act: -0.0002

Data Index: 1040685 (C4 (Web Text))

Max Activating Token Index: 906

Text #9

Max Range: 0.0507. Min Range: -0.0507

Max Act: 0.0324. Min Act: -0.0002

Data Index: 365443 (C4 (Web Text))

Max Activating Token Index: 302

Text #10

Max Range: 0.0507. Min Range: -0.0507

Max Act: 0.0297. Min Act: -0.0002

Data Index: 27425 (C4 (Web Text))

Max Activating Token Index: 62

Text #11

Max Range: 0.0507. Min Range: -0.0507

Max Act: 0.0268. Min Act: -0.0002

Data Index: 997287 (C4 (Web Text))

Max Activating Token Index: 692

Text #12

Max Range: 0.0507. Min Range: -0.0507

Max Act: 0.0306. Min Act: -0.0002

Data Index: 90018 (C4 (Web Text))

Max Activating Token Index: 174

Text #13

Max Range: 0.0507. Min Range: -0.0507

Max Act: 0.0282. Min Act: -0.0002

Data Index: 1584944 (Python Code)

Max Activating Token Index: 225

Text #14

Max Range: 0.0507. Min Range: -0.0507

Max Act: 0.0288. Min Act: -0.0002

Data Index: 598076 (C4 (Web Text))

Max Activating Token Index: 186

Text #15

Max Range: 0.0507. Min Range: -0.0507

Max Act: 0.0264. Min Act: -0.0002

Data Index: 851023 (C4 (Web Text))

Max Activating Token Index: 251

Text #16

Max Range: 0.0507. Min Range: -0.0507

Max Act: 0.0283. Min Act: -0.0002

Data Index: 132307 (C4 (Web Text))

Max Activating Token Index: 263

Text #17

Max Range: 0.0507. Min Range: -0.0507

Max Act: 0.0289. Min Act: -0.0002

Data Index: 348902 (C4 (Web Text))

Max Activating Token Index: 553

Text #18

Max Range: 0.0507. Min Range: -0.0507

Max Act: 0.0280. Min Act: -0.0002

Data Index: 223041 (C4 (Web Text))

Max Activating Token Index: 684

Text #19

Max Range: 0.0507. Min Range: -0.0507

Max Act: 0.0268. Min Act: -0.0002

Data Index: 1105643 (C4 (Web Text))

Max Activating Token Index: 662