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

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

Neuron 2031 in Layer 9

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.6505. Min Range: -0.6505

Max Act: 0.6505. Min Act: -0.0000

Data Index: 323103 (C4 (Web Text))

Max Activating Token Index: 89

Text #1

Max Range: 0.6505. Min Range: -0.6505

Max Act: 0.6361. Min Act: -0.0000

Data Index: 1414872 (Python Code)

Max Activating Token Index: 846

Text #2

Max Range: 0.6505. Min Range: -0.6505

Max Act: 0.6194. Min Act: -0.0000

Data Index: 1061507 (C4 (Web Text))

Max Activating Token Index: 714

Text #3

Max Range: 0.6505. Min Range: -0.6505

Max Act: 0.5232. Min Act: -0.0000

Data Index: 1477765 (Python Code)

Max Activating Token Index: 870

Text #4

Max Range: 0.6505. Min Range: -0.6505

Max Act: 0.5069. Min Act: -0.0000

Data Index: 973810 (C4 (Web Text))

Max Activating Token Index: 504

Text #5

Max Range: 0.6505. Min Range: -0.6505

Max Act: 0.4340. Min Act: 0.0000

Data Index: 1531901 (Python Code)

Max Activating Token Index: 499

Text #6

Max Range: 0.6505. Min Range: -0.6505

Max Act: 0.4476. Min Act: -0.0000

Data Index: 893535 (C4 (Web Text))

Max Activating Token Index: 450

Text #7

Max Range: 0.6505. Min Range: -0.6505

Max Act: 0.4616. Min Act: -0.0000

Data Index: 77850 (C4 (Web Text))

Max Activating Token Index: 467

Text #8

Max Range: 0.6505. Min Range: -0.6505

Max Act: 0.4577. Min Act: -0.0000

Data Index: 105777 (C4 (Web Text))

Max Activating Token Index: 675

Text #9

Max Range: 0.6505. Min Range: -0.6505

Max Act: 0.4425. Min Act: -0.0000

Data Index: 607414 (C4 (Web Text))

Max Activating Token Index: 348

Text #10

Max Range: 0.6505. Min Range: -0.6505

Max Act: 0.5171. Min Act: -0.0000

Data Index: 440306 (C4 (Web Text))

Max Activating Token Index: 224

Text #11

Max Range: 0.6505. Min Range: -0.6505

Max Act: 0.4258. Min Act: -0.0000

Data Index: 1567366 (Python Code)

Max Activating Token Index: 498

Text #12

Max Range: 0.6505. Min Range: -0.6505

Max Act: 0.4479. Min Act: -0.0000

Data Index: 122589 (C4 (Web Text))

Max Activating Token Index: 813

Text #13

Max Range: 0.6505. Min Range: -0.6505

Max Act: 0.3563. Min Act: -0.0000

Data Index: 1646864 (Python Code)

Max Activating Token Index: 237

Text #14

Max Range: 0.6505. Min Range: -0.6505

Max Act: 0.4282. Min Act: -0.0000

Data Index: 47964 (C4 (Web Text))

Max Activating Token Index: 683

Text #15

Max Range: 0.6505. Min Range: -0.6505

Max Act: 0.3843. Min Act: -0.0000

Data Index: 396469 (C4 (Web Text))

Max Activating Token Index: 975

Text #16

Max Range: 0.6505. Min Range: -0.6505

Max Act: 0.4270. Min Act: -0.0000

Data Index: 997124 (C4 (Web Text))

Max Activating Token Index: 688

Text #17

Max Range: 0.6505. Min Range: -0.6505

Max Act: 0.4506. Min Act: -0.0000

Data Index: 1050718 (C4 (Web Text))

Max Activating Token Index: 834

Text #18

Max Range: 0.6505. Min Range: -0.6505

Max Act: 0.3923. Min Act: -0.0000

Data Index: 29488 (C4 (Web Text))

Max Activating Token Index: 664

Text #19

Max Range: 0.6505. Min Range: -0.6505

Max Act: 0.4199. Min Act: -0.0000

Data Index: 777002 (C4 (Web Text))

Max Activating Token Index: 892