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

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

Neuron 398 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: 0.0285. Min Range: -0.0285

Max Act: 0.0271. Min Act: -0.0001

Data Index: 56289 (C4 (Web Text))

Max Activating Token Index: 317

Text #1

Max Range: 0.0285. Min Range: -0.0285

Max Act: 0.0266. Min Act: -0.0001

Data Index: 886026 (C4 (Web Text))

Max Activating Token Index: 895

Text #2

Max Range: 0.0285. Min Range: -0.0285

Max Act: 0.0264. Min Act: -0.0001

Data Index: 905338 (C4 (Web Text))

Max Activating Token Index: 659

Text #3

Max Range: 0.0285. Min Range: -0.0285

Max Act: 0.0285. Min Act: -0.0001

Data Index: 288632 (C4 (Web Text))

Max Activating Token Index: 669

Text #4

Max Range: 0.0285. Min Range: -0.0285

Max Act: 0.0227. Min Act: -0.0001

Data Index: 665954 (C4 (Web Text))

Max Activating Token Index: 785

Text #5

Max Range: 0.0285. Min Range: -0.0285

Max Act: 0.0222. Min Act: -0.0001

Data Index: 869116 (C4 (Web Text))

Max Activating Token Index: 467

Text #6

Max Range: 0.0285. Min Range: -0.0285

Max Act: 0.0219. Min Act: -0.0001

Data Index: 1219809 (C4 (Web Text))

Max Activating Token Index: 750

Text #7

Max Range: 0.0285. Min Range: -0.0285

Max Act: 0.0211. Min Act: -0.0001

Data Index: 286237 (C4 (Web Text))

Max Activating Token Index: 598

Text #8

Max Range: 0.0285. Min Range: -0.0285

Max Act: 0.0232. Min Act: -0.0001

Data Index: 794370 (C4 (Web Text))

Max Activating Token Index: 135

Text #9

Max Range: 0.0285. Min Range: -0.0285

Max Act: 0.0257. Min Act: -0.0001

Data Index: 684425 (C4 (Web Text))

Max Activating Token Index: 871

Text #10

Max Range: 0.0285. Min Range: -0.0285

Max Act: 0.0208. Min Act: -0.0001

Data Index: 619520 (C4 (Web Text))

Max Activating Token Index: 222

Text #11

Max Range: 0.0285. Min Range: -0.0285

Max Act: 0.0212. Min Act: -0.0001

Data Index: 1278591 (C4 (Web Text))

Max Activating Token Index: 29

Text #12

Max Range: 0.0285. Min Range: -0.0285

Max Act: 0.0217. Min Act: -0.0001

Data Index: 1349820 (C4 (Web Text))

Max Activating Token Index: 241

Text #13

Max Range: 0.0285. Min Range: -0.0285

Max Act: 0.0209. Min Act: -0.0001

Data Index: 836718 (C4 (Web Text))

Max Activating Token Index: 397

Text #14

Max Range: 0.0285. Min Range: -0.0285

Max Act: 0.0223. Min Act: -0.0002

Data Index: 1614364 (Python Code)

Max Activating Token Index: 669

Text #15

Max Range: 0.0285. Min Range: -0.0285

Max Act: 0.0209. Min Act: -0.0001

Data Index: 671870 (C4 (Web Text))

Max Activating Token Index: 432

Text #16

Max Range: 0.0285. Min Range: -0.0285

Max Act: 0.0207. Min Act: -0.0001

Data Index: 337703 (C4 (Web Text))

Max Activating Token Index: 431

Text #17

Max Range: 0.0285. Min Range: -0.0285

Max Act: 0.0198. Min Act: -0.0001

Data Index: 883505 (C4 (Web Text))

Max Activating Token Index: 564

Text #18

Max Range: 0.0285. Min Range: -0.0285

Max Act: 0.0196. Min Act: -0.0001

Data Index: 198281 (C4 (Web Text))

Max Activating Token Index: 230

Text #19

Max Range: 0.0285. Min Range: -0.0285

Max Act: 0.0207. Min Act: -0.0001

Data Index: 862898 (C4 (Web Text))

Max Activating Token Index: 630