Model: SoLU Model: 12 Layers, 6144 Neurons per Layer

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

Neuron 4320 in Layer 3

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-12l')

Text #0

Max Range: 0.0414. Min Range: -0.0414

Max Act: 0.0414. Min Act: -0.0001

Data Index: 840594 (C4 (Web Text))

Max Activating Token Index: 466

Text #1

Max Range: 0.0414. Min Range: -0.0414

Max Act: 0.0369. Min Act: -0.0001

Data Index: 364136 (C4 (Web Text))

Max Activating Token Index: 344

Text #2

Max Range: 0.0414. Min Range: -0.0414

Max Act: 0.0264. Min Act: -0.0001

Data Index: 347088 (C4 (Web Text))

Max Activating Token Index: 314

Text #3

Max Range: 0.0414. Min Range: -0.0414

Max Act: 0.0248. Min Act: -0.0001

Data Index: 1116520 (C4 (Web Text))

Max Activating Token Index: 140

Text #4

Max Range: 0.0414. Min Range: -0.0414

Max Act: 0.0237. Min Act: -0.0001

Data Index: 100357 (C4 (Web Text))

Max Activating Token Index: 265

Text #5

Max Range: 0.0414. Min Range: -0.0414

Max Act: 0.0254. Min Act: -0.0001

Data Index: 406812 (C4 (Web Text))

Max Activating Token Index: 577

Text #6

Max Range: 0.0414. Min Range: -0.0414

Max Act: 0.0248. Min Act: -0.0001

Data Index: 1229934 (C4 (Web Text))

Max Activating Token Index: 710

Text #7

Max Range: 0.0414. Min Range: -0.0414

Max Act: 0.0240. Min Act: -0.0001

Data Index: 7249 (C4 (Web Text))

Max Activating Token Index: 687

Text #8

Max Range: 0.0414. Min Range: -0.0414

Max Act: 0.0223. Min Act: -0.0001

Data Index: 425537 (C4 (Web Text))

Max Activating Token Index: 548

Text #9

Max Range: 0.0414. Min Range: -0.0414

Max Act: 0.0193. Min Act: -0.0001

Data Index: 45931 (C4 (Web Text))

Max Activating Token Index: 357

Text #10

Max Range: 0.0414. Min Range: -0.0414

Max Act: 0.0218. Min Act: -0.0001

Data Index: 1609855 (Python Code)

Max Activating Token Index: 3

Text #11

Max Range: 0.0414. Min Range: -0.0414

Max Act: 0.0228. Min Act: -0.0001

Data Index: 1027846 (C4 (Web Text))

Max Activating Token Index: 351

Text #12

Max Range: 0.0414. Min Range: -0.0414

Max Act: 0.0176. Min Act: -0.0001

Data Index: 666043 (C4 (Web Text))

Max Activating Token Index: 267

Text #13

Max Range: 0.0414. Min Range: -0.0414

Max Act: 0.0230. Min Act: -0.0001

Data Index: 292694 (C4 (Web Text))

Max Activating Token Index: 658

Text #14

Max Range: 0.0414. Min Range: -0.0414

Max Act: 0.0214. Min Act: -0.0001

Data Index: 1589503 (Python Code)

Max Activating Token Index: 645

Text #15

Max Range: 0.0414. Min Range: -0.0414

Max Act: 0.0195. Min Act: -0.0001

Data Index: 115883 (C4 (Web Text))

Max Activating Token Index: 716

Text #16

Max Range: 0.0414. Min Range: -0.0414

Max Act: 0.0215. Min Act: -0.0001

Data Index: 382287 (C4 (Web Text))

Max Activating Token Index: 484

Text #17

Max Range: 0.0414. Min Range: -0.0414

Max Act: 0.0213. Min Act: -0.0001

Data Index: 1060898 (C4 (Web Text))

Max Activating Token Index: 714

Text #18

Max Range: 0.0414. Min Range: -0.0414

Max Act: 0.0192. Min Act: -0.0001

Data Index: 208301 (C4 (Web Text))

Max Activating Token Index: 426

Text #19

Max Range: 0.0414. Min Range: -0.0414

Max Act: 0.0177. Min Act: -0.0001

Data Index: 594292 (C4 (Web Text))

Max Activating Token Index: 175