Model: SoLU Model: 10 Layers, 5120 Neurons per Layer

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

Neuron 4300 in Layer 9

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-10l')

Text #0

Max Range: 0.3572. Min Range: -0.3572

Max Act: 0.3572. Min Act: 0.0000

Data Index: 321614 (C4 (Web Text))

Max Activating Token Index: 554

Text #1

Max Range: 0.3572. Min Range: -0.3572

Max Act: 0.3512. Min Act: -0.0000

Data Index: 1195381 (C4 (Web Text))

Max Activating Token Index: 97

Text #2

Max Range: 0.3572. Min Range: -0.3572

Max Act: 0.3310. Min Act: -0.0000

Data Index: 847135 (C4 (Web Text))

Max Activating Token Index: 119

Text #3

Max Range: 0.3572. Min Range: -0.3572

Max Act: 0.3394. Min Act: -0.0000

Data Index: 1392838 (Python Code)

Max Activating Token Index: 847

Text #4

Max Range: 0.3572. Min Range: -0.3572

Max Act: 0.3095. Min Act: -0.0000

Data Index: 781394 (C4 (Web Text))

Max Activating Token Index: 103

Text #5

Max Range: 0.3572. Min Range: -0.3572

Max Act: 0.3007. Min Act: -0.0000

Data Index: 1436718 (Python Code)

Max Activating Token Index: 994

Text #6

Max Range: 0.3572. Min Range: -0.3572

Max Act: 0.3102. Min Act: -0.0000

Data Index: 667389 (C4 (Web Text))

Max Activating Token Index: 662

Text #7

Max Range: 0.3572. Min Range: -0.3572

Max Act: 0.3012. Min Act: 0.0000

Data Index: 708279 (C4 (Web Text))

Max Activating Token Index: 552

Text #8

Max Range: 0.3572. Min Range: -0.3572

Max Act: 0.3084. Min Act: 0.0000

Data Index: 971750 (C4 (Web Text))

Max Activating Token Index: 353

Text #9

Max Range: 0.3572. Min Range: -0.3572

Max Act: 0.2562. Min Act: -0.0000

Data Index: 1613538 (Python Code)

Max Activating Token Index: 703

Text #10

Max Range: 0.3572. Min Range: -0.3572

Max Act: 0.2808. Min Act: -0.0000

Data Index: 328430 (C4 (Web Text))

Max Activating Token Index: 959

Text #11

Max Range: 0.3572. Min Range: -0.3572

Max Act: 0.2597. Min Act: -0.0000

Data Index: 1528456 (Python Code)

Max Activating Token Index: 357

Text #12

Max Range: 0.3572. Min Range: -0.3572

Max Act: 0.2687. Min Act: -0.0000

Data Index: 141749 (C4 (Web Text))

Max Activating Token Index: 672

Text #13

Max Range: 0.3572. Min Range: -0.3572

Max Act: 0.2583. Min Act: -0.0000

Data Index: 796876 (C4 (Web Text))

Max Activating Token Index: 632

Text #14

Max Range: 0.3572. Min Range: -0.3572

Max Act: 0.2570. Min Act: -0.0000

Data Index: 1552458 (Python Code)

Max Activating Token Index: 620

Text #15

Max Range: 0.3572. Min Range: -0.3572

Max Act: 0.2625. Min Act: -0.0000

Data Index: 92496 (C4 (Web Text))

Max Activating Token Index: 747

Text #16

Max Range: 0.3572. Min Range: -0.3572

Max Act: 0.2631. Min Act: -0.0000

Data Index: 534309 (C4 (Web Text))

Max Activating Token Index: 459

Text #17

Max Range: 0.3572. Min Range: -0.3572

Max Act: 0.2803. Min Act: -0.0000

Data Index: 1401228 (Python Code)

Max Activating Token Index: 654

Text #18

Max Range: 0.3572. Min Range: -0.3572

Max Act: 0.2506. Min Act: 0.0000

Data Index: 222288 (C4 (Web Text))

Max Activating Token Index: 988

Text #19

Max Range: 0.3572. Min Range: -0.3572

Max Act: 0.2659. Min Act: 0.0001

Data Index: 181597 (C4 (Web Text))

Max Activating Token Index: 280