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

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

Neuron 97 in Layer 1

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.0250. Min Range: -0.0250

Max Act: 0.0250. Min Act: -0.0004

Data Index: 656768 (C4 (Web Text))

Max Activating Token Index: 355

Text #1

Max Range: 0.0250. Min Range: -0.0250

Max Act: 0.0250. Min Act: -0.0003

Data Index: 280575 (C4 (Web Text))

Max Activating Token Index: 232

Text #2

Max Range: 0.0250. Min Range: -0.0250

Max Act: 0.0225. Min Act: -0.0003

Data Index: 515285 (C4 (Web Text))

Max Activating Token Index: 717

Text #3

Max Range: 0.0250. Min Range: -0.0250

Max Act: 0.0199. Min Act: -0.0003

Data Index: 1486556 (Python Code)

Max Activating Token Index: 717

Text #4

Max Range: 0.0250. Min Range: -0.0250

Max Act: 0.0198. Min Act: -0.0003

Data Index: 1163679 (C4 (Web Text))

Max Activating Token Index: 926

Text #5

Max Range: 0.0250. Min Range: -0.0250

Max Act: 0.0179. Min Act: -0.0003

Data Index: 742013 (C4 (Web Text))

Max Activating Token Index: 769

Text #6

Max Range: 0.0250. Min Range: -0.0250

Max Act: 0.0213. Min Act: -0.0003

Data Index: 1540537 (Python Code)

Max Activating Token Index: 425

Text #7

Max Range: 0.0250. Min Range: -0.0250

Max Act: 0.0193. Min Act: -0.0003

Data Index: 1254688 (C4 (Web Text))

Max Activating Token Index: 607

Text #8

Max Range: 0.0250. Min Range: -0.0250

Max Act: 0.0187. Min Act: -0.0003

Data Index: 607831 (C4 (Web Text))

Max Activating Token Index: 752

Text #9

Max Range: 0.0250. Min Range: -0.0250

Max Act: 0.0193. Min Act: -0.0003

Data Index: 1367651 (Python Code)

Max Activating Token Index: 567

Text #10

Max Range: 0.0250. Min Range: -0.0250

Max Act: 0.0188. Min Act: -0.0003

Data Index: 1334360 (C4 (Web Text))

Max Activating Token Index: 368

Text #11

Max Range: 0.0250. Min Range: -0.0250

Max Act: 0.0192. Min Act: -0.0003

Data Index: 1653339 (Python Code)

Max Activating Token Index: 498

Text #12

Max Range: 0.0250. Min Range: -0.0250

Max Act: 0.0203. Min Act: -0.0003

Data Index: 240967 (C4 (Web Text))

Max Activating Token Index: 526

Text #13

Max Range: 0.0250. Min Range: -0.0250

Max Act: 0.0180. Min Act: -0.0003

Data Index: 1069387 (C4 (Web Text))

Max Activating Token Index: 662

Text #14

Max Range: 0.0250. Min Range: -0.0250

Max Act: 0.0196. Min Act: -0.0003

Data Index: 1520152 (Python Code)

Max Activating Token Index: 120

Text #15

Max Range: 0.0250. Min Range: -0.0250

Max Act: 0.0194. Min Act: -0.0003

Data Index: 1650357 (Python Code)

Max Activating Token Index: 721

Text #16

Max Range: 0.0250. Min Range: -0.0250

Max Act: 0.0186. Min Act: -0.0003

Data Index: 1053892 (C4 (Web Text))

Max Activating Token Index: 504

Text #17

Max Range: 0.0250. Min Range: -0.0250

Max Act: 0.0185. Min Act: -0.0003

Data Index: 253449 (C4 (Web Text))

Max Activating Token Index: 471

Text #18

Max Range: 0.0250. Min Range: -0.0250

Max Act: 0.0185. Min Act: -0.0003

Data Index: 399577 (C4 (Web Text))

Max Activating Token Index: 901

Text #19

Max Range: 0.0250. Min Range: -0.0250

Max Act: 0.0184. Min Act: -0.0003

Data Index: 36949 (C4 (Web Text))

Max Activating Token Index: 748