Model: SoLU Model: 12 Layers, 6144 Neurons per Layer

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

Neuron 4525 in Layer 7

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-12l')

Text #0

Max Range: 0.2708. Min Range: -0.2708

Max Act: 0.2708. Min Act: -0.0000

Data Index: 1553462 (Python Code)

Max Activating Token Index: 626

Text #1

Max Range: 0.2708. Min Range: -0.2708

Max Act: 0.2704. Min Act: -0.0000

Data Index: 464012 (C4 (Web Text))

Max Activating Token Index: 306

Text #2

Max Range: 0.2708. Min Range: -0.2708

Max Act: 0.2509. Min Act: -0.0000

Data Index: 446960 (C4 (Web Text))

Max Activating Token Index: 143

Text #3

Max Range: 0.2708. Min Range: -0.2708

Max Act: 0.2584. Min Act: -0.0000

Data Index: 1372415 (Python Code)

Max Activating Token Index: 608

Text #4

Max Range: 0.2708. Min Range: -0.2708

Max Act: 0.2306. Min Act: -0.0000

Data Index: 277593 (C4 (Web Text))

Max Activating Token Index: 144

Text #5

Max Range: 0.2708. Min Range: -0.2708

Max Act: 0.2479. Min Act: -0.0000

Data Index: 92534 (C4 (Web Text))

Max Activating Token Index: 782

Text #6

Max Range: 0.2708. Min Range: -0.2708

Max Act: 0.2303. Min Act: -0.0000

Data Index: 1363452 (Python Code)

Max Activating Token Index: 889

Text #7

Max Range: 0.2708. Min Range: -0.2708

Max Act: 0.2450. Min Act: -0.0000

Data Index: 592863 (C4 (Web Text))

Max Activating Token Index: 559

Text #8

Max Range: 0.2708. Min Range: -0.2708

Max Act: 0.2432. Min Act: -0.0000

Data Index: 1219081 (C4 (Web Text))

Max Activating Token Index: 391

Text #9

Max Range: 0.2708. Min Range: -0.2708

Max Act: 0.2335. Min Act: -0.0000

Data Index: 311902 (C4 (Web Text))

Max Activating Token Index: 569

Text #10

Max Range: 0.2708. Min Range: -0.2708

Max Act: 0.2175. Min Act: -0.0000

Data Index: 211482 (C4 (Web Text))

Max Activating Token Index: 180

Text #11

Max Range: 0.2708. Min Range: -0.2708

Max Act: 0.2081. Min Act: -0.0000

Data Index: 1650634 (Python Code)

Max Activating Token Index: 869

Text #12

Max Range: 0.2708. Min Range: -0.2708

Max Act: 0.2330. Min Act: -0.0000

Data Index: 556642 (C4 (Web Text))

Max Activating Token Index: 348

Text #13

Max Range: 0.2708. Min Range: -0.2708

Max Act: 0.2007. Min Act: -0.0000

Data Index: 1306453 (C4 (Web Text))

Max Activating Token Index: 714

Text #14

Max Range: 0.2708. Min Range: -0.2708

Max Act: 0.2205. Min Act: -0.0000

Data Index: 1303366 (C4 (Web Text))

Max Activating Token Index: 650

Text #15

Max Range: 0.2708. Min Range: -0.2708

Max Act: 0.2190. Min Act: -0.0000

Data Index: 1053243 (C4 (Web Text))

Max Activating Token Index: 233

Text #16

Max Range: 0.2708. Min Range: -0.2708

Max Act: 0.2291. Min Act: -0.0000

Data Index: 1344619 (C4 (Web Text))

Max Activating Token Index: 372

Text #17

Max Range: 0.2708. Min Range: -0.2708

Max Act: 0.2125. Min Act: -0.0000

Data Index: 1373389 (Python Code)

Max Activating Token Index: 962

Text #18

Max Range: 0.2708. Min Range: -0.2708

Max Act: 0.2166. Min Act: -0.0000

Data Index: 1122931 (C4 (Web Text))

Max Activating Token Index: 272

Text #19

Max Range: 0.2708. Min Range: -0.2708

Max Act: 0.1832. Min Act: -0.0000

Data Index: 175488 (C4 (Web Text))

Max Activating Token Index: 684