Model: SoLU Model: 3 Layers, 2048 Neurons per Layer

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

Neuron 194 in Layer 1

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-3l')

Text #0

Max Range: 0.0397. Min Range: -0.0397

Max Act: 0.0397. Min Act: -0.0006

Data Index: 884850 (C4 (Web Text))

Max Activating Token Index: 155

Text #1

Max Range: 0.0397. Min Range: -0.0397

Max Act: 0.0393. Min Act: -0.0006

Data Index: 1078311 (C4 (Web Text))

Max Activating Token Index: 343

Text #2

Max Range: 0.0397. Min Range: -0.0397

Max Act: 0.0346. Min Act: -0.0006

Data Index: 618745 (C4 (Web Text))

Max Activating Token Index: 524

Text #3

Max Range: 0.0397. Min Range: -0.0397

Max Act: 0.0325. Min Act: -0.0006

Data Index: 765549 (C4 (Web Text))

Max Activating Token Index: 621

Text #4

Max Range: 0.0397. Min Range: -0.0397

Max Act: 0.0335. Min Act: -0.0006

Data Index: 105849 (C4 (Web Text))

Max Activating Token Index: 1021

Text #5

Max Range: 0.0397. Min Range: -0.0397

Max Act: 0.0313. Min Act: -0.0006

Data Index: 1187295 (C4 (Web Text))

Max Activating Token Index: 338

Text #6

Max Range: 0.0397. Min Range: -0.0397

Max Act: 0.0320. Min Act: -0.0006

Data Index: 1116808 (C4 (Web Text))

Max Activating Token Index: 624

Text #7

Max Range: 0.0397. Min Range: -0.0397

Max Act: 0.0323. Min Act: -0.0006

Data Index: 323851 (C4 (Web Text))

Max Activating Token Index: 373

Text #8

Max Range: 0.0397. Min Range: -0.0397

Max Act: 0.0323. Min Act: -0.0005

Data Index: 978422 (C4 (Web Text))

Max Activating Token Index: 934

Text #9

Max Range: 0.0397. Min Range: -0.0397

Max Act: 0.0300. Min Act: -0.0006

Data Index: 254715 (C4 (Web Text))

Max Activating Token Index: 391

Text #10

Max Range: 0.0397. Min Range: -0.0397

Max Act: 0.0306. Min Act: -0.0006

Data Index: 939670 (C4 (Web Text))

Max Activating Token Index: 306

Text #11

Max Range: 0.0397. Min Range: -0.0397

Max Act: 0.0299. Min Act: -0.0006

Data Index: 255305 (C4 (Web Text))

Max Activating Token Index: 722

Text #12

Max Range: 0.0397. Min Range: -0.0397

Max Act: 0.0309. Min Act: -0.0006

Data Index: 125102 (C4 (Web Text))

Max Activating Token Index: 991

Text #13

Max Range: 0.0397. Min Range: -0.0397

Max Act: 0.0316. Min Act: -0.0006

Data Index: 963507 (C4 (Web Text))

Max Activating Token Index: 237

Text #14

Max Range: 0.0397. Min Range: -0.0397

Max Act: 0.0303. Min Act: -0.0006

Data Index: 1312067 (C4 (Web Text))

Max Activating Token Index: 261

Text #15

Max Range: 0.0397. Min Range: -0.0397

Max Act: 0.0309. Min Act: -0.0006

Data Index: 1622166 (Python Code)

Max Activating Token Index: 956

Text #16

Max Range: 0.0397. Min Range: -0.0397

Max Act: 0.0306. Min Act: -0.0006

Data Index: 1500624 (Python Code)

Max Activating Token Index: 331

Text #17

Max Range: 0.0397. Min Range: -0.0397

Max Act: 0.0318. Min Act: -0.0006

Data Index: 568415 (C4 (Web Text))

Max Activating Token Index: 330

Text #18

Max Range: 0.0397. Min Range: -0.0397

Max Act: 0.0305. Min Act: -0.0006

Data Index: 1002201 (C4 (Web Text))

Max Activating Token Index: 935

Text #19

Max Range: 0.0397. Min Range: -0.0397

Max Act: 0.0308. Min Act: -0.0006

Data Index: 551146 (C4 (Web Text))

Max Activating Token Index: 506