Model: SoLU Model: 8 Layers, 4096 Neurons per Layer

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

Neuron 3420 in Layer 3

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-8l')

Text #0

Max Range: 0.0293. Min Range: -0.0293

Max Act: 0.0293. Min Act: -0.0001

Data Index: 752769 (C4 (Web Text))

Max Activating Token Index: 124

Text #1

Max Range: 0.0293. Min Range: -0.0293

Max Act: 0.0280. Min Act: -0.0001

Data Index: 727666 (C4 (Web Text))

Max Activating Token Index: 892

Text #2

Max Range: 0.0293. Min Range: -0.0293

Max Act: 0.0270. Min Act: -0.0001

Data Index: 431007 (C4 (Web Text))

Max Activating Token Index: 420

Text #3

Max Range: 0.0293. Min Range: -0.0293

Max Act: 0.0273. Min Act: -0.0001

Data Index: 94175 (C4 (Web Text))

Max Activating Token Index: 835

Text #4

Max Range: 0.0293. Min Range: -0.0293

Max Act: 0.0253. Min Act: -0.0001

Data Index: 1483637 (Python Code)

Max Activating Token Index: 255

Text #5

Max Range: 0.0293. Min Range: -0.0293

Max Act: 0.0240. Min Act: -0.0001

Data Index: 323783 (C4 (Web Text))

Max Activating Token Index: 472

Text #6

Max Range: 0.0293. Min Range: -0.0293

Max Act: 0.0244. Min Act: -0.0001

Data Index: 1413396 (Python Code)

Max Activating Token Index: 301

Text #7

Max Range: 0.0293. Min Range: -0.0293

Max Act: 0.0227. Min Act: -0.0001

Data Index: 1000793 (C4 (Web Text))

Max Activating Token Index: 919

Text #8

Max Range: 0.0293. Min Range: -0.0293

Max Act: 0.0229. Min Act: -0.0001

Data Index: 845687 (C4 (Web Text))

Max Activating Token Index: 446

Text #9

Max Range: 0.0293. Min Range: -0.0293

Max Act: 0.0238. Min Act: -0.0001

Data Index: 763338 (C4 (Web Text))

Max Activating Token Index: 63

Text #10

Max Range: 0.0293. Min Range: -0.0293

Max Act: 0.0209. Min Act: -0.0001

Data Index: 1345100 (C4 (Web Text))

Max Activating Token Index: 532

Text #11

Max Range: 0.0293. Min Range: -0.0293

Max Act: 0.0236. Min Act: -0.0001

Data Index: 1469528 (Python Code)

Max Activating Token Index: 544

Text #12

Max Range: 0.0293. Min Range: -0.0293

Max Act: 0.0242. Min Act: -0.0001

Data Index: 925489 (C4 (Web Text))

Max Activating Token Index: 125

Text #13

Max Range: 0.0293. Min Range: -0.0293

Max Act: 0.0240. Min Act: -0.0001

Data Index: 1605371 (Python Code)

Max Activating Token Index: 971

Text #14

Max Range: 0.0293. Min Range: -0.0293

Max Act: 0.0222. Min Act: -0.0001

Data Index: 198512 (C4 (Web Text))

Max Activating Token Index: 392

Text #15

Max Range: 0.0293. Min Range: -0.0293

Max Act: 0.0215. Min Act: -0.0001

Data Index: 47770 (C4 (Web Text))

Max Activating Token Index: 517

Text #16

Max Range: 0.0293. Min Range: -0.0293

Max Act: 0.0225. Min Act: -0.0001

Data Index: 10473 (C4 (Web Text))

Max Activating Token Index: 543

Text #17

Max Range: 0.0293. Min Range: -0.0293

Max Act: 0.0220. Min Act: -0.0001

Data Index: 1593729 (Python Code)

Max Activating Token Index: 207

Text #18

Max Range: 0.0293. Min Range: -0.0293

Max Act: 0.0224. Min Act: -0.0001

Data Index: 1288107 (C4 (Web Text))

Max Activating Token Index: 178

Text #19

Max Range: 0.0293. Min Range: -0.0293

Max Act: 0.0192. Min Act: -0.0001

Data Index: 1320948 (C4 (Web Text))

Max Activating Token Index: 865