Model: SoLU Model: 4 Layers, 2048 Neurons per Layer

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

Neuron 18 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-4l')

Text #0

Max Range: 0.0310. Min Range: -0.0310

Max Act: 0.0310. Min Act: -0.0005

Data Index: 394060 (C4 (Web Text))

Max Activating Token Index: 125

Text #1

Max Range: 0.0310. Min Range: -0.0310

Max Act: 0.0287. Min Act: -0.0005

Data Index: 42382 (C4 (Web Text))

Max Activating Token Index: 152

Text #2

Max Range: 0.0310. Min Range: -0.0310

Max Act: 0.0280. Min Act: -0.0005

Data Index: 1174630 (C4 (Web Text))

Max Activating Token Index: 20

Text #3

Max Range: 0.0310. Min Range: -0.0310

Max Act: 0.0288. Min Act: -0.0005

Data Index: 498306 (C4 (Web Text))

Max Activating Token Index: 407

Text #4

Max Range: 0.0310. Min Range: -0.0310

Max Act: 0.0270. Min Act: -0.0005

Data Index: 106893 (C4 (Web Text))

Max Activating Token Index: 77

Text #5

Max Range: 0.0310. Min Range: -0.0310

Max Act: 0.0274. Min Act: -0.0005

Data Index: 1178163 (C4 (Web Text))

Max Activating Token Index: 22

Text #6

Max Range: 0.0310. Min Range: -0.0310

Max Act: 0.0273. Min Act: -0.0005

Data Index: 701993 (C4 (Web Text))

Max Activating Token Index: 40

Text #7

Max Range: 0.0310. Min Range: -0.0310

Max Act: 0.0253. Min Act: -0.0005

Data Index: 986098 (C4 (Web Text))

Max Activating Token Index: 180

Text #8

Max Range: 0.0310. Min Range: -0.0310

Max Act: 0.0261. Min Act: -0.0005

Data Index: 1637994 (Python Code)

Max Activating Token Index: 451

Text #9

Max Range: 0.0310. Min Range: -0.0310

Max Act: 0.0257. Min Act: -0.0005

Data Index: 47866 (C4 (Web Text))

Max Activating Token Index: 91

Text #10

Max Range: 0.0310. Min Range: -0.0310

Max Act: 0.0259. Min Act: -0.0005

Data Index: 412814 (C4 (Web Text))

Max Activating Token Index: 147

Text #11

Max Range: 0.0310. Min Range: -0.0310

Max Act: 0.0251. Min Act: -0.0005

Data Index: 859589 (C4 (Web Text))

Max Activating Token Index: 94

Text #12

Max Range: 0.0310. Min Range: -0.0310

Max Act: 0.0261. Min Act: -0.0005

Data Index: 233282 (C4 (Web Text))

Max Activating Token Index: 92

Text #13

Max Range: 0.0310. Min Range: -0.0310

Max Act: 0.0266. Min Act: -0.0005

Data Index: 131554 (C4 (Web Text))

Max Activating Token Index: 473

Text #14

Max Range: 0.0310. Min Range: -0.0310

Max Act: 0.0244. Min Act: -0.0005

Data Index: 136089 (C4 (Web Text))

Max Activating Token Index: 80

Text #15

Max Range: 0.0310. Min Range: -0.0310

Max Act: 0.0258. Min Act: -0.0005

Data Index: 1545607 (Python Code)

Max Activating Token Index: 593

Text #16

Max Range: 0.0310. Min Range: -0.0310

Max Act: 0.0268. Min Act: -0.0005

Data Index: 1031904 (C4 (Web Text))

Max Activating Token Index: 979

Text #17

Max Range: 0.0310. Min Range: -0.0310

Max Act: 0.0262. Min Act: -0.0005

Data Index: 457244 (C4 (Web Text))

Max Activating Token Index: 100

Text #18

Max Range: 0.0310. Min Range: -0.0310

Max Act: 0.0254. Min Act: -0.0005

Data Index: 1548923 (Python Code)

Max Activating Token Index: 678

Text #19

Max Range: 0.0310. Min Range: -0.0310

Max Act: 0.0236. Min Act: -0.0005

Data Index: 479183 (C4 (Web Text))

Max Activating Token Index: 193