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

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

Neuron 4350 in Layer 4

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.4748. Min Range: -0.4748

Max Act: 0.4748. Min Act: -0.0001

Data Index: 524767 (C4 (Web Text))

Max Activating Token Index: 631

Text #1

Max Range: 0.4748. Min Range: -0.4748

Max Act: 0.4486. Min Act: -0.0001

Data Index: 367508 (C4 (Web Text))

Max Activating Token Index: 550

Text #2

Max Range: 0.4748. Min Range: -0.4748

Max Act: 0.4497. Min Act: -0.0000

Data Index: 714094 (C4 (Web Text))

Max Activating Token Index: 727

Text #3

Max Range: 0.4748. Min Range: -0.4748

Max Act: 0.4509. Min Act: -0.0000

Data Index: 93797 (C4 (Web Text))

Max Activating Token Index: 629

Text #4

Max Range: 0.4748. Min Range: -0.4748

Max Act: 0.4530. Min Act: -0.0001

Data Index: 193306 (C4 (Web Text))

Max Activating Token Index: 219

Text #5

Max Range: 0.4748. Min Range: -0.4748

Max Act: 0.4329. Min Act: -0.0000

Data Index: 1648555 (Python Code)

Max Activating Token Index: 757

Text #6

Max Range: 0.4748. Min Range: -0.4748

Max Act: 0.4279. Min Act: -0.0001

Data Index: 486230 (C4 (Web Text))

Max Activating Token Index: 494

Text #7

Max Range: 0.4748. Min Range: -0.4748

Max Act: 0.4635. Min Act: -0.0000

Data Index: 1407398 (Python Code)

Max Activating Token Index: 577

Text #8

Max Range: 0.4748. Min Range: -0.4748

Max Act: 0.4275. Min Act: -0.0000

Data Index: 1469114 (Python Code)

Max Activating Token Index: 543

Text #9

Max Range: 0.4748. Min Range: -0.4748

Max Act: 0.3506. Min Act: -0.0001

Data Index: 1182300 (C4 (Web Text))

Max Activating Token Index: 343

Text #10

Max Range: 0.4748. Min Range: -0.4748

Max Act: 0.3486. Min Act: -0.0001

Data Index: 686485 (C4 (Web Text))

Max Activating Token Index: 755

Text #11

Max Range: 0.4748. Min Range: -0.4748

Max Act: 0.3951. Min Act: -0.0001

Data Index: 691900 (C4 (Web Text))

Max Activating Token Index: 388

Text #12

Max Range: 0.4748. Min Range: -0.4748

Max Act: 0.3888. Min Act: -0.0001

Data Index: 800183 (C4 (Web Text))

Max Activating Token Index: 41

Text #13

Max Range: 0.4748. Min Range: -0.4748

Max Act: 0.3936. Min Act: -0.0000

Data Index: 1303363 (C4 (Web Text))

Max Activating Token Index: 292

Text #14

Max Range: 0.4748. Min Range: -0.4748

Max Act: 0.3667. Min Act: -0.0001

Data Index: 996701 (C4 (Web Text))

Max Activating Token Index: 802

Text #15

Max Range: 0.4748. Min Range: -0.4748

Max Act: 0.3543. Min Act: -0.0000

Data Index: 1402849 (Python Code)

Max Activating Token Index: 150

Text #16

Max Range: 0.4748. Min Range: -0.4748

Max Act: 0.3180. Min Act: -0.0000

Data Index: 120810 (C4 (Web Text))

Max Activating Token Index: 718

Text #17

Max Range: 0.4748. Min Range: -0.4748

Max Act: 0.3894. Min Act: -0.0000

Data Index: 757450 (C4 (Web Text))

Max Activating Token Index: 231

Text #18

Max Range: 0.4748. Min Range: -0.4748

Max Act: 0.3676. Min Act: -0.0001

Data Index: 15365 (C4 (Web Text))

Max Activating Token Index: 616

Text #19

Max Range: 0.4748. Min Range: -0.4748

Max Act: 0.3682. Min Act: -0.0000

Data Index: 1546211 (Python Code)

Max Activating Token Index: 976