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

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

Neuron 351 in Layer 1

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.0233. Min Range: -0.0233

Max Act: 0.0233. Min Act: -0.0006

Data Index: 653777 (C4 (Web Text))

Max Activating Token Index: 614

Text #1

Max Range: 0.0233. Min Range: -0.0233

Max Act: 0.0225. Min Act: -0.0006

Data Index: 569061 (C4 (Web Text))

Max Activating Token Index: 588

Text #2

Max Range: 0.0233. Min Range: -0.0233

Max Act: 0.0218. Min Act: -0.0006

Data Index: 485858 (C4 (Web Text))

Max Activating Token Index: 440

Text #3

Max Range: 0.0233. Min Range: -0.0233

Max Act: 0.0218. Min Act: -0.0006

Data Index: 582473 (C4 (Web Text))

Max Activating Token Index: 118

Text #4

Max Range: 0.0233. Min Range: -0.0233

Max Act: 0.0202. Min Act: -0.0006

Data Index: 1263952 (C4 (Web Text))

Max Activating Token Index: 687

Text #5

Max Range: 0.0233. Min Range: -0.0233

Max Act: 0.0200. Min Act: -0.0006

Data Index: 699884 (C4 (Web Text))

Max Activating Token Index: 800

Text #6

Max Range: 0.0233. Min Range: -0.0233

Max Act: 0.0199. Min Act: -0.0006

Data Index: 243545 (C4 (Web Text))

Max Activating Token Index: 913

Text #7

Max Range: 0.0233. Min Range: -0.0233

Max Act: 0.0203. Min Act: -0.0006

Data Index: 1193841 (C4 (Web Text))

Max Activating Token Index: 731

Text #8

Max Range: 0.0233. Min Range: -0.0233

Max Act: 0.0204. Min Act: -0.0006

Data Index: 563236 (C4 (Web Text))

Max Activating Token Index: 398

Text #9

Max Range: 0.0233. Min Range: -0.0233

Max Act: 0.0212. Min Act: -0.0006

Data Index: 28863 (C4 (Web Text))

Max Activating Token Index: 435

Text #10

Max Range: 0.0233. Min Range: -0.0233

Max Act: 0.0197. Min Act: -0.0006

Data Index: 552992 (C4 (Web Text))

Max Activating Token Index: 957

Text #11

Max Range: 0.0233. Min Range: -0.0233

Max Act: 0.0190. Min Act: -0.0006

Data Index: 1150867 (C4 (Web Text))

Max Activating Token Index: 981

Text #12

Max Range: 0.0233. Min Range: -0.0233

Max Act: 0.0197. Min Act: -0.0006

Data Index: 153524 (C4 (Web Text))

Max Activating Token Index: 621

Text #13

Max Range: 0.0233. Min Range: -0.0233

Max Act: 0.0197. Min Act: -0.0006

Data Index: 10682 (C4 (Web Text))

Max Activating Token Index: 200

Text #14

Max Range: 0.0233. Min Range: -0.0233

Max Act: 0.0187. Min Act: -0.0006

Data Index: 1313554 (C4 (Web Text))

Max Activating Token Index: 708

Text #15

Max Range: 0.0233. Min Range: -0.0233

Max Act: 0.0209. Min Act: -0.0006

Data Index: 243794 (C4 (Web Text))

Max Activating Token Index: 628

Text #16

Max Range: 0.0233. Min Range: -0.0233

Max Act: 0.0200. Min Act: -0.0006

Data Index: 1417056 (Python Code)

Max Activating Token Index: 261

Text #17

Max Range: 0.0233. Min Range: -0.0233

Max Act: 0.0195. Min Act: -0.0006

Data Index: 1112422 (C4 (Web Text))

Max Activating Token Index: 865

Text #18

Max Range: 0.0233. Min Range: -0.0233

Max Act: 0.0199. Min Act: -0.0006

Data Index: 640702 (C4 (Web Text))

Max Activating Token Index: 609

Text #19

Max Range: 0.0233. Min Range: -0.0233

Max Act: 0.0198. Min Act: -0.0006

Data Index: 1237124 (C4 (Web Text))

Max Activating Token Index: 855