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

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

Neuron 54 in Layer 5

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.0636. Min Range: -0.0636

Max Act: 0.0627. Min Act: -0.0001

Data Index: 246863 (C4 (Web Text))

Max Activating Token Index: 895

Text #1

Max Range: 0.0636. Min Range: -0.0636

Max Act: 0.0636. Min Act: -0.0001

Data Index: 1415188 (Python Code)

Max Activating Token Index: 973

Text #2

Max Range: 0.0636. Min Range: -0.0636

Max Act: 0.0597. Min Act: -0.0001

Data Index: 685980 (C4 (Web Text))

Max Activating Token Index: 868

Text #3

Max Range: 0.0636. Min Range: -0.0636

Max Act: 0.0596. Min Act: -0.0001

Data Index: 861736 (C4 (Web Text))

Max Activating Token Index: 137

Text #4

Max Range: 0.0636. Min Range: -0.0636

Max Act: 0.0580. Min Act: -0.0001

Data Index: 1305382 (C4 (Web Text))

Max Activating Token Index: 393

Text #5

Max Range: 0.0636. Min Range: -0.0636

Max Act: 0.0575. Min Act: -0.0001

Data Index: 1343716 (C4 (Web Text))

Max Activating Token Index: 34

Text #6

Max Range: 0.0636. Min Range: -0.0636

Max Act: 0.0573. Min Act: -0.0001

Data Index: 297274 (C4 (Web Text))

Max Activating Token Index: 268

Text #7

Max Range: 0.0636. Min Range: -0.0636

Max Act: 0.0568. Min Act: 0.0000

Data Index: 1194969 (C4 (Web Text))

Max Activating Token Index: 310

Text #8

Max Range: 0.0636. Min Range: -0.0636

Max Act: 0.0584. Min Act: -0.0001

Data Index: 1100979 (C4 (Web Text))

Max Activating Token Index: 89

Text #9

Max Range: 0.0636. Min Range: -0.0636

Max Act: 0.0551. Min Act: -0.0001

Data Index: 1101131 (C4 (Web Text))

Max Activating Token Index: 877

Text #10

Max Range: 0.0636. Min Range: -0.0636

Max Act: 0.0586. Min Act: -0.0001

Data Index: 84992 (C4 (Web Text))

Max Activating Token Index: 54

Text #11

Max Range: 0.0636. Min Range: -0.0636

Max Act: 0.0570. Min Act: -0.0000

Data Index: 1454089 (Python Code)

Max Activating Token Index: 257

Text #12

Max Range: 0.0636. Min Range: -0.0636

Max Act: 0.0577. Min Act: -0.0001

Data Index: 460543 (C4 (Web Text))

Max Activating Token Index: 877

Text #13

Max Range: 0.0636. Min Range: -0.0636

Max Act: 0.0576. Min Act: -0.0001

Data Index: 190845 (C4 (Web Text))

Max Activating Token Index: 58

Text #14

Max Range: 0.0636. Min Range: -0.0636

Max Act: 0.0578. Min Act: -0.0001

Data Index: 1088387 (C4 (Web Text))

Max Activating Token Index: 415

Text #15

Max Range: 0.0636. Min Range: -0.0636

Max Act: 0.0517. Min Act: -0.0001

Data Index: 1589330 (Python Code)

Max Activating Token Index: 767

Text #16

Max Range: 0.0636. Min Range: -0.0636

Max Act: 0.0542. Min Act: -0.0001

Data Index: 274319 (C4 (Web Text))

Max Activating Token Index: 898

Text #17

Max Range: 0.0636. Min Range: -0.0636

Max Act: 0.0555. Min Act: -0.0001

Data Index: 384586 (C4 (Web Text))

Max Activating Token Index: 634

Text #18

Max Range: 0.0636. Min Range: -0.0636

Max Act: 0.0573. Min Act: -0.0001

Data Index: 1526647 (Python Code)

Max Activating Token Index: 641

Text #19

Max Range: 0.0636. Min Range: -0.0636

Max Act: 0.0567. Min Act: -0.0001

Data Index: 1107081 (C4 (Web Text))

Max Activating Token Index: 350