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

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

Neuron 647 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.0407. Min Range: -0.0407

Max Act: 0.0407. Min Act: -0.0001

Data Index: 264449 (C4 (Web Text))

Max Activating Token Index: 375

Text #1

Max Range: 0.0407. Min Range: -0.0407

Max Act: 0.0396. Min Act: -0.0001

Data Index: 761762 (C4 (Web Text))

Max Activating Token Index: 193

Text #2

Max Range: 0.0407. Min Range: -0.0407

Max Act: 0.0390. Min Act: -0.0001

Data Index: 120574 (C4 (Web Text))

Max Activating Token Index: 438

Text #3

Max Range: 0.0407. Min Range: -0.0407

Max Act: 0.0388. Min Act: -0.0001

Data Index: 67614 (C4 (Web Text))

Max Activating Token Index: 882

Text #4

Max Range: 0.0407. Min Range: -0.0407

Max Act: 0.0340. Min Act: -0.0001

Data Index: 982418 (C4 (Web Text))

Max Activating Token Index: 331

Text #5

Max Range: 0.0407. Min Range: -0.0407

Max Act: 0.0329. Min Act: -0.0002

Data Index: 112438 (C4 (Web Text))

Max Activating Token Index: 937

Text #6

Max Range: 0.0407. Min Range: -0.0407

Max Act: 0.0333. Min Act: -0.0001

Data Index: 836554 (C4 (Web Text))

Max Activating Token Index: 71

Text #7

Max Range: 0.0407. Min Range: -0.0407

Max Act: 0.0240. Min Act: -0.0001

Data Index: 1319740 (C4 (Web Text))

Max Activating Token Index: 334

Text #8

Max Range: 0.0407. Min Range: -0.0407

Max Act: 0.0240. Min Act: -0.0001

Data Index: 567905 (C4 (Web Text))

Max Activating Token Index: 597

Text #9

Max Range: 0.0407. Min Range: -0.0407

Max Act: 0.0242. Min Act: -0.0001

Data Index: 1570312 (Python Code)

Max Activating Token Index: 637

Text #10

Max Range: 0.0407. Min Range: -0.0407

Max Act: 0.0235. Min Act: -0.0001

Data Index: 1106528 (C4 (Web Text))

Max Activating Token Index: 311

Text #11

Max Range: 0.0407. Min Range: -0.0407

Max Act: 0.0229. Min Act: -0.0002

Data Index: 99409 (C4 (Web Text))

Max Activating Token Index: 613

Text #12

Max Range: 0.0407. Min Range: -0.0407

Max Act: 0.0256. Min Act: -0.0001

Data Index: 1471150 (Python Code)

Max Activating Token Index: 409

Text #13

Max Range: 0.0407. Min Range: -0.0407

Max Act: 0.0244. Min Act: -0.0001

Data Index: 562287 (C4 (Web Text))

Max Activating Token Index: 130

Text #14

Max Range: 0.0407. Min Range: -0.0407

Max Act: 0.0218. Min Act: -0.0001

Data Index: 1478753 (Python Code)

Max Activating Token Index: 113

Text #15

Max Range: 0.0407. Min Range: -0.0407

Max Act: 0.0209. Min Act: -0.0001

Data Index: 394166 (C4 (Web Text))

Max Activating Token Index: 247

Text #16

Max Range: 0.0407. Min Range: -0.0407

Max Act: 0.0243. Min Act: -0.0001

Data Index: 1166507 (C4 (Web Text))

Max Activating Token Index: 430

Text #17

Max Range: 0.0407. Min Range: -0.0407

Max Act: 0.0223. Min Act: -0.0001

Data Index: 1420347 (Python Code)

Max Activating Token Index: 733

Text #18

Max Range: 0.0407. Min Range: -0.0407

Max Act: 0.0239. Min Act: -0.0001

Data Index: 1105794 (C4 (Web Text))

Max Activating Token Index: 109

Text #19

Max Range: 0.0407. Min Range: -0.0407

Max Act: 0.0209. Min Act: -0.0001

Data Index: 566983 (C4 (Web Text))

Max Activating Token Index: 1022