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

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

Neuron 208 in Layer 3

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.0905. Min Range: -0.0905

Max Act: 0.0905. Min Act: -0.0002

Data Index: 461585 (C4 (Web Text))

Max Activating Token Index: 243

Text #1

Max Range: 0.0905. Min Range: -0.0905

Max Act: 0.0888. Min Act: -0.0002

Data Index: 40324 (C4 (Web Text))

Max Activating Token Index: 364

Text #2

Max Range: 0.0905. Min Range: -0.0905

Max Act: 0.0849. Min Act: -0.0002

Data Index: 1070186 (C4 (Web Text))

Max Activating Token Index: 729

Text #3

Max Range: 0.0905. Min Range: -0.0905

Max Act: 0.0860. Min Act: -0.0002

Data Index: 606570 (C4 (Web Text))

Max Activating Token Index: 825

Text #4

Max Range: 0.0905. Min Range: -0.0905

Max Act: 0.0827. Min Act: -0.0002

Data Index: 583732 (C4 (Web Text))

Max Activating Token Index: 686

Text #5

Max Range: 0.0905. Min Range: -0.0905

Max Act: 0.0835. Min Act: -0.0002

Data Index: 1343333 (C4 (Web Text))

Max Activating Token Index: 45

Text #6

Max Range: 0.0905. Min Range: -0.0905

Max Act: 0.0847. Min Act: -0.0002

Data Index: 1087463 (C4 (Web Text))

Max Activating Token Index: 28

Text #7

Max Range: 0.0905. Min Range: -0.0905

Max Act: 0.0820. Min Act: -0.0002

Data Index: 1563570 (Python Code)

Max Activating Token Index: 281

Text #8

Max Range: 0.0905. Min Range: -0.0905

Max Act: 0.0841. Min Act: -0.0002

Data Index: 1384544 (Python Code)

Max Activating Token Index: 655

Text #9

Max Range: 0.0905. Min Range: -0.0905

Max Act: 0.0823. Min Act: -0.0002

Data Index: 1102413 (C4 (Web Text))

Max Activating Token Index: 727

Text #10

Max Range: 0.0905. Min Range: -0.0905

Max Act: 0.0802. Min Act: -0.0002

Data Index: 1577347 (Python Code)

Max Activating Token Index: 930

Text #11

Max Range: 0.0905. Min Range: -0.0905

Max Act: 0.0814. Min Act: -0.0002

Data Index: 1375543 (Python Code)

Max Activating Token Index: 359

Text #12

Max Range: 0.0905. Min Range: -0.0905

Max Act: 0.0798. Min Act: -0.0002

Data Index: 589318 (C4 (Web Text))

Max Activating Token Index: 23

Text #13

Max Range: 0.0905. Min Range: -0.0905

Max Act: 0.0785. Min Act: -0.0002

Data Index: 1632464 (Python Code)

Max Activating Token Index: 43

Text #14

Max Range: 0.0905. Min Range: -0.0905

Max Act: 0.0783. Min Act: -0.0002

Data Index: 364226 (C4 (Web Text))

Max Activating Token Index: 935

Text #15

Max Range: 0.0905. Min Range: -0.0905

Max Act: 0.0800. Min Act: -0.0002

Data Index: 518328 (C4 (Web Text))

Max Activating Token Index: 800

Text #16

Max Range: 0.0905. Min Range: -0.0905

Max Act: 0.0779. Min Act: -0.0002

Data Index: 1483745 (Python Code)

Max Activating Token Index: 352

Text #17

Max Range: 0.0905. Min Range: -0.0905

Max Act: 0.0773. Min Act: -0.0002

Data Index: 312075 (C4 (Web Text))

Max Activating Token Index: 917

Text #18

Max Range: 0.0905. Min Range: -0.0905

Max Act: 0.0775. Min Act: -0.0002

Data Index: 753951 (C4 (Web Text))

Max Activating Token Index: 754

Text #19

Max Range: 0.0905. Min Range: -0.0905

Max Act: 0.0763. Min Act: -0.0002

Data Index: 895190 (C4 (Web Text))

Max Activating Token Index: 669