Model: SoLU Model: 8 Layers, 4096 Neurons per Layer

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

Neuron 2139 in Layer 4

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-8l')

Text #0

Max Range: 0.5534. Min Range: -0.5534

Max Act: 0.5534. Min Act: -0.0001

Data Index: 144039 (C4 (Web Text))

Max Activating Token Index: 596

Text #1

Max Range: 0.5534. Min Range: -0.5534

Max Act: 0.5524. Min Act: -0.0001

Data Index: 49958 (C4 (Web Text))

Max Activating Token Index: 357

Text #2

Max Range: 0.5534. Min Range: -0.5534

Max Act: 0.5343. Min Act: -0.0001

Data Index: 887342 (C4 (Web Text))

Max Activating Token Index: 787

Text #3

Max Range: 0.5534. Min Range: -0.5534

Max Act: 0.5111. Min Act: -0.0001

Data Index: 498040 (C4 (Web Text))

Max Activating Token Index: 546

Text #4

Max Range: 0.5534. Min Range: -0.5534

Max Act: 0.4930. Min Act: -0.0001

Data Index: 437927 (C4 (Web Text))

Max Activating Token Index: 538

Text #5

Max Range: 0.5534. Min Range: -0.5534

Max Act: 0.4902. Min Act: -0.0001

Data Index: 147862 (C4 (Web Text))

Max Activating Token Index: 992

Text #6

Max Range: 0.5534. Min Range: -0.5534

Max Act: 0.4716. Min Act: -0.0001

Data Index: 1023852 (C4 (Web Text))

Max Activating Token Index: 71

Text #7

Max Range: 0.5534. Min Range: -0.5534

Max Act: 0.4869. Min Act: -0.0001

Data Index: 1345357 (C4 (Web Text))

Max Activating Token Index: 810

Text #8

Max Range: 0.5534. Min Range: -0.5534

Max Act: 0.5158. Min Act: -0.0001

Data Index: 1136462 (C4 (Web Text))

Max Activating Token Index: 83

Text #9

Max Range: 0.5534. Min Range: -0.5534

Max Act: 0.4930. Min Act: -0.0001

Data Index: 862947 (C4 (Web Text))

Max Activating Token Index: 566

Text #10

Max Range: 0.5534. Min Range: -0.5534

Max Act: 0.4752. Min Act: -0.0001

Data Index: 1612128 (Python Code)

Max Activating Token Index: 605

Text #11

Max Range: 0.5534. Min Range: -0.5534

Max Act: 0.4802. Min Act: -0.0001

Data Index: 931654 (C4 (Web Text))

Max Activating Token Index: 1004

Text #12

Max Range: 0.5534. Min Range: -0.5534

Max Act: 0.4752. Min Act: -0.0001

Data Index: 1038197 (C4 (Web Text))

Max Activating Token Index: 210

Text #13

Max Range: 0.5534. Min Range: -0.5534

Max Act: 0.4578. Min Act: -0.0001

Data Index: 213160 (C4 (Web Text))

Max Activating Token Index: 881

Text #14

Max Range: 0.5534. Min Range: -0.5534

Max Act: 0.4575. Min Act: -0.0001

Data Index: 1000085 (C4 (Web Text))

Max Activating Token Index: 740

Text #15

Max Range: 0.5534. Min Range: -0.5534

Max Act: 0.4665. Min Act: -0.0001

Data Index: 1107827 (C4 (Web Text))

Max Activating Token Index: 513

Text #16

Max Range: 0.5534. Min Range: -0.5534

Max Act: 0.4713. Min Act: -0.0001

Data Index: 1244643 (C4 (Web Text))

Max Activating Token Index: 1000

Text #17

Max Range: 0.5534. Min Range: -0.5534

Max Act: 0.4927. Min Act: -0.0001

Data Index: 380739 (C4 (Web Text))

Max Activating Token Index: 250

Text #18

Max Range: 0.5534. Min Range: -0.5534

Max Act: 0.4729. Min Act: -0.0001

Data Index: 1573625 (Python Code)

Max Activating Token Index: 667

Text #19

Max Range: 0.5534. Min Range: -0.5534

Max Act: 0.4564. Min Act: -0.0001

Data Index: 148783 (C4 (Web Text))

Max Activating Token Index: 167