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

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

Neuron 4844 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: 0.0710. Min Range: -0.0710

Max Act: 0.0689. Min Act: -0.0001

Data Index: 88664 (C4 (Web Text))

Max Activating Token Index: 159

Text #1

Max Range: 0.0710. Min Range: -0.0710

Max Act: 0.0710. Min Act: -0.0001

Data Index: 13884 (C4 (Web Text))

Max Activating Token Index: 250

Text #2

Max Range: 0.0710. Min Range: -0.0710

Max Act: 0.0667. Min Act: -0.0001

Data Index: 899241 (C4 (Web Text))

Max Activating Token Index: 120

Text #3

Max Range: 0.0710. Min Range: -0.0710

Max Act: 0.0691. Min Act: -0.0001

Data Index: 568365 (C4 (Web Text))

Max Activating Token Index: 340

Text #4

Max Range: 0.0710. Min Range: -0.0710

Max Act: 0.0637. Min Act: -0.0001

Data Index: 684816 (C4 (Web Text))

Max Activating Token Index: 825

Text #5

Max Range: 0.0710. Min Range: -0.0710

Max Act: 0.0639. Min Act: -0.0001

Data Index: 1307613 (C4 (Web Text))

Max Activating Token Index: 700

Text #6

Max Range: 0.0710. Min Range: -0.0710

Max Act: 0.0644. Min Act: -0.0001

Data Index: 330321 (C4 (Web Text))

Max Activating Token Index: 618

Text #7

Max Range: 0.0710. Min Range: -0.0710

Max Act: 0.0629. Min Act: -0.0001

Data Index: 1020029 (C4 (Web Text))

Max Activating Token Index: 940

Text #8

Max Range: 0.0710. Min Range: -0.0710

Max Act: 0.0619. Min Act: -0.0001

Data Index: 689685 (C4 (Web Text))

Max Activating Token Index: 76

Text #9

Max Range: 0.0710. Min Range: -0.0710

Max Act: 0.0592. Min Act: -0.0001

Data Index: 334667 (C4 (Web Text))

Max Activating Token Index: 389

Text #10

Max Range: 0.0710. Min Range: -0.0710

Max Act: 0.0585. Min Act: -0.0001

Data Index: 1061230 (C4 (Web Text))

Max Activating Token Index: 1022

Text #11

Max Range: 0.0710. Min Range: -0.0710

Max Act: 0.0585. Min Act: -0.0001

Data Index: 783267 (C4 (Web Text))

Max Activating Token Index: 298

Text #12

Max Range: 0.0710. Min Range: -0.0710

Max Act: 0.0604. Min Act: -0.0001

Data Index: 555062 (C4 (Web Text))

Max Activating Token Index: 507

Text #13

Max Range: 0.0710. Min Range: -0.0710

Max Act: 0.0602. Min Act: -0.0001

Data Index: 546968 (C4 (Web Text))

Max Activating Token Index: 778

Text #14

Max Range: 0.0710. Min Range: -0.0710

Max Act: 0.0566. Min Act: -0.0001

Data Index: 1303245 (C4 (Web Text))

Max Activating Token Index: 625

Text #15

Max Range: 0.0710. Min Range: -0.0710

Max Act: 0.0595. Min Act: -0.0001

Data Index: 11256 (C4 (Web Text))

Max Activating Token Index: 382

Text #16

Max Range: 0.0710. Min Range: -0.0710

Max Act: 0.0574. Min Act: -0.0001

Data Index: 55822 (C4 (Web Text))

Max Activating Token Index: 115

Text #17

Max Range: 0.0710. Min Range: -0.0710

Max Act: 0.0554. Min Act: -0.0001

Data Index: 1361243 (Python Code)

Max Activating Token Index: 60

Text #18

Max Range: 0.0710. Min Range: -0.0710

Max Act: 0.0575. Min Act: -0.0001

Data Index: 491649 (C4 (Web Text))

Max Activating Token Index: 570

Text #19

Max Range: 0.0710. Min Range: -0.0710

Max Act: 0.0603. Min Act: -0.0001

Data Index: 553747 (C4 (Web Text))

Max Activating Token Index: 588