Model: SoLU Model: 6 Layers, 3072 Neurons per Layer

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

Neuron 2004 in Layer 4

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-6l')

Text #0

Max Range: 0.0227. Min Range: -0.0227

Max Act: 0.0227. Min Act: -0.0001

Data Index: 101787 (C4 (Web Text))

Max Activating Token Index: 296

Text #1

Max Range: 0.0227. Min Range: -0.0227

Max Act: 0.0203. Min Act: -0.0001

Data Index: 559054 (C4 (Web Text))

Max Activating Token Index: 343

Text #2

Max Range: 0.0227. Min Range: -0.0227

Max Act: 0.0205. Min Act: -0.0001

Data Index: 373882 (C4 (Web Text))

Max Activating Token Index: 353

Text #3

Max Range: 0.0227. Min Range: -0.0227

Max Act: 0.0187. Min Act: -0.0001

Data Index: 987580 (C4 (Web Text))

Max Activating Token Index: 1000

Text #4

Max Range: 0.0227. Min Range: -0.0227

Max Act: 0.0183. Min Act: -0.0001

Data Index: 919589 (C4 (Web Text))

Max Activating Token Index: 369

Text #5

Max Range: 0.0227. Min Range: -0.0227

Max Act: 0.0194. Min Act: -0.0001

Data Index: 668276 (C4 (Web Text))

Max Activating Token Index: 258

Text #6

Max Range: 0.0227. Min Range: -0.0227

Max Act: 0.0182. Min Act: -0.0001

Data Index: 1249286 (C4 (Web Text))

Max Activating Token Index: 805

Text #7

Max Range: 0.0227. Min Range: -0.0227

Max Act: 0.0184. Min Act: -0.0001

Data Index: 1070690 (C4 (Web Text))

Max Activating Token Index: 421

Text #8

Max Range: 0.0227. Min Range: -0.0227

Max Act: 0.0182. Min Act: -0.0001

Data Index: 1304082 (C4 (Web Text))

Max Activating Token Index: 906

Text #9

Max Range: 0.0227. Min Range: -0.0227

Max Act: 0.0183. Min Act: -0.0001

Data Index: 862073 (C4 (Web Text))

Max Activating Token Index: 342

Text #10

Max Range: 0.0227. Min Range: -0.0227

Max Act: 0.0180. Min Act: -0.0001

Data Index: 239592 (C4 (Web Text))

Max Activating Token Index: 447

Text #11

Max Range: 0.0227. Min Range: -0.0227

Max Act: 0.0177. Min Act: -0.0001

Data Index: 1244162 (C4 (Web Text))

Max Activating Token Index: 868

Text #12

Max Range: 0.0227. Min Range: -0.0227

Max Act: 0.0171. Min Act: -0.0001

Data Index: 1565427 (Python Code)

Max Activating Token Index: 278

Text #13

Max Range: 0.0227. Min Range: -0.0227

Max Act: 0.0171. Min Act: -0.0001

Data Index: 1336484 (C4 (Web Text))

Max Activating Token Index: 370

Text #14

Max Range: 0.0227. Min Range: -0.0227

Max Act: 0.0166. Min Act: -0.0001

Data Index: 553106 (C4 (Web Text))

Max Activating Token Index: 69

Text #15

Max Range: 0.0227. Min Range: -0.0227

Max Act: 0.0167. Min Act: -0.0001

Data Index: 881266 (C4 (Web Text))

Max Activating Token Index: 431

Text #16

Max Range: 0.0227. Min Range: -0.0227

Max Act: 0.0165. Min Act: -0.0001

Data Index: 1349220 (C4 (Web Text))

Max Activating Token Index: 365

Text #17

Max Range: 0.0227. Min Range: -0.0227

Max Act: 0.0164. Min Act: -0.0001

Data Index: 235196 (C4 (Web Text))

Max Activating Token Index: 936

Text #18

Max Range: 0.0227. Min Range: -0.0227

Max Act: 0.0158. Min Act: -0.0001

Data Index: 120399 (C4 (Web Text))

Max Activating Token Index: 801

Text #19

Max Range: 0.0227. Min Range: -0.0227

Max Act: 0.0163. Min Act: -0.0001

Data Index: 790463 (C4 (Web Text))

Max Activating Token Index: 855