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

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

Neuron 1297 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: 0.0275. Min Range: -0.0275

Max Act: 0.0275. Min Act: -0.0003

Data Index: 1451783 (Python Code)

Max Activating Token Index: 910

Text #1

Max Range: 0.0275. Min Range: -0.0275

Max Act: 0.0266. Min Act: -0.0003

Data Index: 588074 (C4 (Web Text))

Max Activating Token Index: 192

Text #2

Max Range: 0.0275. Min Range: -0.0275

Max Act: 0.0233. Min Act: -0.0003

Data Index: 758270 (C4 (Web Text))

Max Activating Token Index: 1015

Text #3

Max Range: 0.0275. Min Range: -0.0275

Max Act: 0.0246. Min Act: -0.0003

Data Index: 964303 (C4 (Web Text))

Max Activating Token Index: 22

Text #4

Max Range: 0.0275. Min Range: -0.0275

Max Act: 0.0244. Min Act: -0.0003

Data Index: 1641193 (Python Code)

Max Activating Token Index: 866

Text #5

Max Range: 0.0275. Min Range: -0.0275

Max Act: 0.0236. Min Act: -0.0003

Data Index: 764668 (C4 (Web Text))

Max Activating Token Index: 192

Text #6

Max Range: 0.0275. Min Range: -0.0275

Max Act: 0.0232. Min Act: -0.0003

Data Index: 1462803 (Python Code)

Max Activating Token Index: 649

Text #7

Max Range: 0.0275. Min Range: -0.0275

Max Act: 0.0233. Min Act: -0.0002

Data Index: 1053835 (C4 (Web Text))

Max Activating Token Index: 595

Text #8

Max Range: 0.0275. Min Range: -0.0275

Max Act: 0.0223. Min Act: -0.0003

Data Index: 344302 (C4 (Web Text))

Max Activating Token Index: 995

Text #9

Max Range: 0.0275. Min Range: -0.0275

Max Act: 0.0231. Min Act: -0.0003

Data Index: 414664 (C4 (Web Text))

Max Activating Token Index: 725

Text #10

Max Range: 0.0275. Min Range: -0.0275

Max Act: 0.0225. Min Act: -0.0002

Data Index: 88755 (C4 (Web Text))

Max Activating Token Index: 29

Text #11

Max Range: 0.0275. Min Range: -0.0275

Max Act: 0.0224. Min Act: -0.0003

Data Index: 1083012 (C4 (Web Text))

Max Activating Token Index: 240

Text #12

Max Range: 0.0275. Min Range: -0.0275

Max Act: 0.0225. Min Act: -0.0002

Data Index: 1296024 (C4 (Web Text))

Max Activating Token Index: 990

Text #13

Max Range: 0.0275. Min Range: -0.0275

Max Act: 0.0225. Min Act: -0.0003

Data Index: 106960 (C4 (Web Text))

Max Activating Token Index: 33

Text #14

Max Range: 0.0275. Min Range: -0.0275

Max Act: 0.0223. Min Act: -0.0002

Data Index: 578901 (C4 (Web Text))

Max Activating Token Index: 511

Text #15

Max Range: 0.0275. Min Range: -0.0275

Max Act: 0.0225. Min Act: -0.0003

Data Index: 1383680 (Python Code)

Max Activating Token Index: 910

Text #16

Max Range: 0.0275. Min Range: -0.0275

Max Act: 0.0225. Min Act: -0.0003

Data Index: 453565 (C4 (Web Text))

Max Activating Token Index: 449

Text #17

Max Range: 0.0275. Min Range: -0.0275

Max Act: 0.0216. Min Act: -0.0003

Data Index: 137 (C4 (Web Text))

Max Activating Token Index: 411

Text #18

Max Range: 0.0275. Min Range: -0.0275

Max Act: 0.0220. Min Act: -0.0003

Data Index: 1337554 (C4 (Web Text))

Max Activating Token Index: 257

Text #19

Max Range: 0.0275. Min Range: -0.0275

Max Act: 0.0213. Min Act: -0.0003

Data Index: 938030 (C4 (Web Text))

Max Activating Token Index: 96