Model: SoLU Model: 12 Layers, 6144 Neurons per Layer

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

Neuron 4782 in Layer 10

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-12l')

Text #0

Max Range: 4.2450. Min Range: -4.2450

Max Act: 4.2450. Min Act: -0.0000

Data Index: 1193710 (C4 (Web Text))

Max Activating Token Index: 440

Text #1

Max Range: 4.2450. Min Range: -4.2450

Max Act: 3.9136. Min Act: -0.0000

Data Index: 975567 (C4 (Web Text))

Max Activating Token Index: 532

Text #2

Max Range: 4.2450. Min Range: -4.2450

Max Act: 3.2054. Min Act: -0.0000

Data Index: 982298 (C4 (Web Text))

Max Activating Token Index: 634

Text #3

Max Range: 4.2450. Min Range: -4.2450

Max Act: 3.0624. Min Act: 0.0000

Data Index: 1116712 (C4 (Web Text))

Max Activating Token Index: 347

Text #4

Max Range: 4.2450. Min Range: -4.2450

Max Act: 2.8777. Min Act: -0.0000

Data Index: 294881 (C4 (Web Text))

Max Activating Token Index: 140

Text #5

Max Range: 4.2450. Min Range: -4.2450

Max Act: 2.8214. Min Act: 0.0000

Data Index: 337691 (C4 (Web Text))

Max Activating Token Index: 33

Text #6

Max Range: 4.2450. Min Range: -4.2450

Max Act: 2.7318. Min Act: -0.0000

Data Index: 253270 (C4 (Web Text))

Max Activating Token Index: 462

Text #7

Max Range: 4.2450. Min Range: -4.2450

Max Act: 2.7255. Min Act: 0.0000

Data Index: 462115 (C4 (Web Text))

Max Activating Token Index: 426

Text #8

Max Range: 4.2450. Min Range: -4.2450

Max Act: 2.5409. Min Act: -0.0000

Data Index: 1443310 (Python Code)

Max Activating Token Index: 671

Text #9

Max Range: 4.2450. Min Range: -4.2450

Max Act: 2.2843. Min Act: -0.0000

Data Index: 346656 (C4 (Web Text))

Max Activating Token Index: 76

Text #10

Max Range: 4.2450. Min Range: -4.2450

Max Act: 2.2611. Min Act: -0.0000

Data Index: 907271 (C4 (Web Text))

Max Activating Token Index: 518

Text #11

Max Range: 4.2450. Min Range: -4.2450

Max Act: 2.1608. Min Act: -0.0000

Data Index: 1647082 (Python Code)

Max Activating Token Index: 90

Text #12

Max Range: 4.2450. Min Range: -4.2450

Max Act: 2.1522. Min Act: 0.0000

Data Index: 203400 (C4 (Web Text))

Max Activating Token Index: 1010

Text #13

Max Range: 4.2450. Min Range: -4.2450

Max Act: 2.1510. Min Act: -0.0000

Data Index: 1080845 (C4 (Web Text))

Max Activating Token Index: 537

Text #14

Max Range: 4.2450. Min Range: -4.2450

Max Act: 2.1270. Min Act: -0.0000

Data Index: 1642729 (Python Code)

Max Activating Token Index: 530

Text #15

Max Range: 4.2450. Min Range: -4.2450

Max Act: 2.0122. Min Act: 0.0001

Data Index: 1075581 (C4 (Web Text))

Max Activating Token Index: 787

Text #16

Max Range: 4.2450. Min Range: -4.2450

Max Act: 2.0868. Min Act: 0.0000

Data Index: 1309966 (C4 (Web Text))

Max Activating Token Index: 524

Text #17

Max Range: 4.2450. Min Range: -4.2450

Max Act: 2.0845. Min Act: 0.0000

Data Index: 977449 (C4 (Web Text))

Max Activating Token Index: 482

Text #18

Max Range: 4.2450. Min Range: -4.2450

Max Act: 2.0709. Min Act: 0.0000

Data Index: 1612789 (Python Code)

Max Activating Token Index: 44

Text #19

Max Range: 4.2450. Min Range: -4.2450

Max Act: 1.9061. Min Act: -0.0000

Data Index: 680822 (C4 (Web Text))

Max Activating Token Index: 473