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

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

Neuron 1630 in Layer 4

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.1084. Min Range: -0.1084

Max Act: 0.0789. Min Act: -0.0001

Data Index: 1188974 (C4 (Web Text))

Max Activating Token Index: 294

Text #1

Max Range: 0.1084. Min Range: -0.1084

Max Act: 0.1084. Min Act: -0.0000

Data Index: 112284 (C4 (Web Text))

Max Activating Token Index: 278

Text #2

Max Range: 0.1084. Min Range: -0.1084

Max Act: 0.0772. Min Act: -0.0001

Data Index: 177036 (C4 (Web Text))

Max Activating Token Index: 589

Text #3

Max Range: 0.1084. Min Range: -0.1084

Max Act: 0.0868. Min Act: -0.0001

Data Index: 598960 (C4 (Web Text))

Max Activating Token Index: 782

Text #4

Max Range: 0.1084. Min Range: -0.1084

Max Act: 0.0708. Min Act: -0.0001

Data Index: 1247510 (C4 (Web Text))

Max Activating Token Index: 37

Text #5

Max Range: 0.1084. Min Range: -0.1084

Max Act: 0.0664. Min Act: -0.0000

Data Index: 662866 (C4 (Web Text))

Max Activating Token Index: 3

Text #6

Max Range: 0.1084. Min Range: -0.1084

Max Act: 0.0664. Min Act: -0.0001

Data Index: 340302 (C4 (Web Text))

Max Activating Token Index: 3

Text #7

Max Range: 0.1084. Min Range: -0.1084

Max Act: 0.0639. Min Act: -0.0001

Data Index: 311464 (C4 (Web Text))

Max Activating Token Index: 32

Text #8

Max Range: 0.1084. Min Range: -0.1084

Max Act: 0.0667. Min Act: -0.0000

Data Index: 318883 (C4 (Web Text))

Max Activating Token Index: 969

Text #9

Max Range: 0.1084. Min Range: -0.1084

Max Act: 0.0405. Min Act: -0.0001

Data Index: 440841 (C4 (Web Text))

Max Activating Token Index: 868

Text #10

Max Range: 0.1084. Min Range: -0.1084

Max Act: 0.0553. Min Act: -0.0000

Data Index: 97690 (C4 (Web Text))

Max Activating Token Index: 385

Text #11

Max Range: 0.1084. Min Range: -0.1084

Max Act: 0.0503. Min Act: -0.0000

Data Index: 202886 (C4 (Web Text))

Max Activating Token Index: 1013

Text #12

Max Range: 0.1084. Min Range: -0.1084

Max Act: 0.0493. Min Act: -0.0000

Data Index: 1246837 (C4 (Web Text))

Max Activating Token Index: 714

Text #13

Max Range: 0.1084. Min Range: -0.1084

Max Act: 0.0562. Min Act: -0.0001

Data Index: 316647 (C4 (Web Text))

Max Activating Token Index: 381

Text #14

Max Range: 0.1084. Min Range: -0.1084

Max Act: 0.0519. Min Act: -0.0000

Data Index: 1563234 (Python Code)

Max Activating Token Index: 173

Text #15

Max Range: 0.1084. Min Range: -0.1084

Max Act: 0.0491. Min Act: -0.0000

Data Index: 1484072 (Python Code)

Max Activating Token Index: 97

Text #16

Max Range: 0.1084. Min Range: -0.1084

Max Act: 0.0514. Min Act: -0.0000

Data Index: 1401177 (Python Code)

Max Activating Token Index: 136

Text #17

Max Range: 0.1084. Min Range: -0.1084

Max Act: 0.0521. Min Act: -0.0001

Data Index: 356557 (C4 (Web Text))

Max Activating Token Index: 42

Text #18

Max Range: 0.1084. Min Range: -0.1084

Max Act: 0.0322. Min Act: -0.0000

Data Index: 1104989 (C4 (Web Text))

Max Activating Token Index: 112

Text #19

Max Range: 0.1084. Min Range: -0.1084

Max Act: 0.0453. Min Act: -0.0001

Data Index: 108971 (C4 (Web Text))

Max Activating Token Index: 6