Model: SoLU Model: 6 Layers, 3072 Neurons per Layer

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

Neuron 2512 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-6l')

Text #0

Max Range: 0.0979. Min Range: -0.0979

Max Act: 0.0849. Min Act: -0.0003

Data Index: 65286 (C4 (Web Text))

Max Activating Token Index: 364

Text #1

Max Range: 0.0979. Min Range: -0.0979

Max Act: 0.0842. Min Act: -0.0003

Data Index: 427832 (C4 (Web Text))

Max Activating Token Index: 200

Text #2

Max Range: 0.0979. Min Range: -0.0979

Max Act: 0.0758. Min Act: -0.0003

Data Index: 526762 (C4 (Web Text))

Max Activating Token Index: 911

Text #3

Max Range: 0.0979. Min Range: -0.0979

Max Act: 0.0762. Min Act: -0.0003

Data Index: 426186 (C4 (Web Text))

Max Activating Token Index: 435

Text #4

Max Range: 0.0979. Min Range: -0.0979

Max Act: 0.0744. Min Act: -0.0003

Data Index: 394581 (C4 (Web Text))

Max Activating Token Index: 210

Text #5

Max Range: 0.0979. Min Range: -0.0979

Max Act: 0.0979. Min Act: -0.0003

Data Index: 975553 (C4 (Web Text))

Max Activating Token Index: 862

Text #6

Max Range: 0.0979. Min Range: -0.0979

Max Act: 0.0839. Min Act: -0.0003

Data Index: 678891 (C4 (Web Text))

Max Activating Token Index: 718

Text #7

Max Range: 0.0979. Min Range: -0.0979

Max Act: 0.0775. Min Act: -0.0003

Data Index: 236611 (C4 (Web Text))

Max Activating Token Index: 28

Text #8

Max Range: 0.0979. Min Range: -0.0979

Max Act: 0.0908. Min Act: -0.0003

Data Index: 1086690 (C4 (Web Text))

Max Activating Token Index: 745

Text #9

Max Range: 0.0979. Min Range: -0.0979

Max Act: 0.0746. Min Act: -0.0003

Data Index: 458982 (C4 (Web Text))

Max Activating Token Index: 84

Text #10

Max Range: 0.0979. Min Range: -0.0979

Max Act: 0.0820. Min Act: -0.0003

Data Index: 300942 (C4 (Web Text))

Max Activating Token Index: 367

Text #11

Max Range: 0.0979. Min Range: -0.0979

Max Act: 0.0612. Min Act: -0.0003

Data Index: 10391 (C4 (Web Text))

Max Activating Token Index: 986

Text #12

Max Range: 0.0979. Min Range: -0.0979

Max Act: 0.0693. Min Act: -0.0003

Data Index: 295895 (C4 (Web Text))

Max Activating Token Index: 33

Text #13

Max Range: 0.0979. Min Range: -0.0979

Max Act: 0.0558. Min Act: -0.0003

Data Index: 482080 (C4 (Web Text))

Max Activating Token Index: 940

Text #14

Max Range: 0.0979. Min Range: -0.0979

Max Act: 0.0678. Min Act: -0.0003

Data Index: 1022087 (C4 (Web Text))

Max Activating Token Index: 176

Text #15

Max Range: 0.0979. Min Range: -0.0979

Max Act: 0.0567. Min Act: -0.0003

Data Index: 597644 (C4 (Web Text))

Max Activating Token Index: 617

Text #16

Max Range: 0.0979. Min Range: -0.0979

Max Act: 0.0555. Min Act: -0.0003

Data Index: 1580789 (Python Code)

Max Activating Token Index: 843

Text #17

Max Range: 0.0979. Min Range: -0.0979

Max Act: 0.0655. Min Act: -0.0003

Data Index: 319181 (C4 (Web Text))

Max Activating Token Index: 570

Text #18

Max Range: 0.0979. Min Range: -0.0979

Max Act: 0.0641. Min Act: -0.0003

Data Index: 429941 (C4 (Web Text))

Max Activating Token Index: 4

Text #19

Max Range: 0.0979. Min Range: -0.0979

Max Act: 0.0816. Min Act: -0.0003

Data Index: 1200986 (C4 (Web Text))

Max Activating Token Index: 874