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

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

Neuron 3780 in Layer 7

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.6398. Min Range: -0.6398

Max Act: 0.6398. Min Act: -0.0000

Data Index: 31971 (C4 (Web Text))

Max Activating Token Index: 135

Text #1

Max Range: 0.6398. Min Range: -0.6398

Max Act: 0.5363. Min Act: -0.0000

Data Index: 446377 (C4 (Web Text))

Max Activating Token Index: 544

Text #2

Max Range: 0.6398. Min Range: -0.6398

Max Act: 0.4936. Min Act: 0.0000

Data Index: 474478 (C4 (Web Text))

Max Activating Token Index: 544

Text #3

Max Range: 0.6398. Min Range: -0.6398

Max Act: 0.5405. Min Act: -0.0000

Data Index: 330637 (C4 (Web Text))

Max Activating Token Index: 316

Text #4

Max Range: 0.6398. Min Range: -0.6398

Max Act: 0.4921. Min Act: -0.0000

Data Index: 89009 (C4 (Web Text))

Max Activating Token Index: 212

Text #5

Max Range: 0.6398. Min Range: -0.6398

Max Act: 0.4798. Min Act: -0.0000

Data Index: 702752 (C4 (Web Text))

Max Activating Token Index: 33

Text #6

Max Range: 0.6398. Min Range: -0.6398

Max Act: 0.4839. Min Act: -0.0000

Data Index: 328531 (C4 (Web Text))

Max Activating Token Index: 80

Text #7

Max Range: 0.6398. Min Range: -0.6398

Max Act: 0.4606. Min Act: -0.0000

Data Index: 1240235 (C4 (Web Text))

Max Activating Token Index: 303

Text #8

Max Range: 0.6398. Min Range: -0.6398

Max Act: 0.4152. Min Act: -0.0000

Data Index: 931990 (C4 (Web Text))

Max Activating Token Index: 372

Text #9

Max Range: 0.6398. Min Range: -0.6398

Max Act: 0.4001. Min Act: -0.0000

Data Index: 1641755 (Python Code)

Max Activating Token Index: 704

Text #10

Max Range: 0.6398. Min Range: -0.6398

Max Act: 0.3984. Min Act: -0.0000

Data Index: 1153845 (C4 (Web Text))

Max Activating Token Index: 11

Text #11

Max Range: 0.6398. Min Range: -0.6398

Max Act: 0.3731. Min Act: -0.0000

Data Index: 1153647 (C4 (Web Text))

Max Activating Token Index: 815

Text #12

Max Range: 0.6398. Min Range: -0.6398

Max Act: 0.3812. Min Act: -0.0000

Data Index: 1378527 (Python Code)

Max Activating Token Index: 651

Text #13

Max Range: 0.6398. Min Range: -0.6398

Max Act: 0.3883. Min Act: -0.0000

Data Index: 738982 (C4 (Web Text))

Max Activating Token Index: 682

Text #14

Max Range: 0.6398. Min Range: -0.6398

Max Act: 0.3800. Min Act: -0.0000

Data Index: 431538 (C4 (Web Text))

Max Activating Token Index: 384

Text #15

Max Range: 0.6398. Min Range: -0.6398

Max Act: 0.3745. Min Act: -0.0000

Data Index: 325357 (C4 (Web Text))

Max Activating Token Index: 482

Text #16

Max Range: 0.6398. Min Range: -0.6398

Max Act: 0.3739. Min Act: -0.0000

Data Index: 1320083 (C4 (Web Text))

Max Activating Token Index: 834

Text #17

Max Range: 0.6398. Min Range: -0.6398

Max Act: 0.3505. Min Act: -0.0000

Data Index: 377598 (C4 (Web Text))

Max Activating Token Index: 579

Text #18

Max Range: 0.6398. Min Range: -0.6398

Max Act: 0.3601. Min Act: -0.0000

Data Index: 482407 (C4 (Web Text))

Max Activating Token Index: 228

Text #19

Max Range: 0.6398. Min Range: -0.6398

Max Act: 0.3106. Min Act: -0.0000

Data Index: 265579 (C4 (Web Text))

Max Activating Token Index: 233