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

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

Neuron 3571 in Layer 8

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.5981. Min Range: -0.5981

Max Act: 0.5981. Min Act: -0.0000

Data Index: 669552 (C4 (Web Text))

Max Activating Token Index: 488

Text #1

Max Range: 0.5981. Min Range: -0.5981

Max Act: 0.5702. Min Act: -0.0000

Data Index: 53459 (C4 (Web Text))

Max Activating Token Index: 474

Text #2

Max Range: 0.5981. Min Range: -0.5981

Max Act: 0.5662. Min Act: -0.0000

Data Index: 1337944 (C4 (Web Text))

Max Activating Token Index: 851

Text #3

Max Range: 0.5981. Min Range: -0.5981

Max Act: 0.5288. Min Act: -0.0000

Data Index: 295488 (C4 (Web Text))

Max Activating Token Index: 652

Text #4

Max Range: 0.5981. Min Range: -0.5981

Max Act: 0.5091. Min Act: -0.0000

Data Index: 1320977 (C4 (Web Text))

Max Activating Token Index: 854

Text #5

Max Range: 0.5981. Min Range: -0.5981

Max Act: 0.4751. Min Act: -0.0000

Data Index: 83861 (C4 (Web Text))

Max Activating Token Index: 815

Text #6

Max Range: 0.5981. Min Range: -0.5981

Max Act: 0.4387. Min Act: -0.0000

Data Index: 1440437 (Python Code)

Max Activating Token Index: 605

Text #7

Max Range: 0.5981. Min Range: -0.5981

Max Act: 0.4472. Min Act: -0.0000

Data Index: 1599255 (Python Code)

Max Activating Token Index: 27

Text #8

Max Range: 0.5981. Min Range: -0.5981

Max Act: 0.4394. Min Act: -0.0000

Data Index: 194064 (C4 (Web Text))

Max Activating Token Index: 142

Text #9

Max Range: 0.5981. Min Range: -0.5981

Max Act: 0.4420. Min Act: -0.0000

Data Index: 1038019 (C4 (Web Text))

Max Activating Token Index: 708

Text #10

Max Range: 0.5981. Min Range: -0.5981

Max Act: 0.4290. Min Act: -0.0000

Data Index: 113829 (C4 (Web Text))

Max Activating Token Index: 491

Text #11

Max Range: 0.5981. Min Range: -0.5981

Max Act: 0.3784. Min Act: -0.0000

Data Index: 880362 (C4 (Web Text))

Max Activating Token Index: 722

Text #12

Max Range: 0.5981. Min Range: -0.5981

Max Act: 0.3632. Min Act: -0.0000

Data Index: 794613 (C4 (Web Text))

Max Activating Token Index: 786

Text #13

Max Range: 0.5981. Min Range: -0.5981

Max Act: 0.3845. Min Act: -0.0000

Data Index: 1438048 (Python Code)

Max Activating Token Index: 815

Text #14

Max Range: 0.5981. Min Range: -0.5981

Max Act: 0.3750. Min Act: -0.0000

Data Index: 1244536 (C4 (Web Text))

Max Activating Token Index: 733

Text #15

Max Range: 0.5981. Min Range: -0.5981

Max Act: 0.3571. Min Act: -0.0000

Data Index: 1631918 (Python Code)

Max Activating Token Index: 142

Text #16

Max Range: 0.5981. Min Range: -0.5981

Max Act: 0.3393. Min Act: -0.0000

Data Index: 560291 (C4 (Web Text))

Max Activating Token Index: 318

Text #17

Max Range: 0.5981. Min Range: -0.5981

Max Act: 0.3411. Min Act: -0.0000

Data Index: 1395791 (Python Code)

Max Activating Token Index: 913

Text #18

Max Range: 0.5981. Min Range: -0.5981

Max Act: 0.3401. Min Act: -0.0000

Data Index: 1057050 (C4 (Web Text))

Max Activating Token Index: 1002

Text #19

Max Range: 0.5981. Min Range: -0.5981

Max Act: 0.3593. Min Act: -0.0000

Data Index: 1070232 (C4 (Web Text))

Max Activating Token Index: 594