Model: SoLU Model: 4 Layers, 2048 Neurons per Layer

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

Neuron 88 in Layer 3

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-4l')

Text #0

Max Range: 0.1231. Min Range: -0.1231

Max Act: 0.1231. Min Act: -0.0002

Data Index: 115921 (C4 (Web Text))

Max Activating Token Index: 200

Text #1

Max Range: 0.1231. Min Range: -0.1231

Max Act: 0.1135. Min Act: -0.0002

Data Index: 243046 (C4 (Web Text))

Max Activating Token Index: 114

Text #2

Max Range: 0.1231. Min Range: -0.1231

Max Act: 0.1099. Min Act: -0.0002

Data Index: 871576 (C4 (Web Text))

Max Activating Token Index: 64

Text #3

Max Range: 0.1231. Min Range: -0.1231

Max Act: 0.1188. Min Act: -0.0002

Data Index: 153057 (C4 (Web Text))

Max Activating Token Index: 780

Text #4

Max Range: 0.1231. Min Range: -0.1231

Max Act: 0.1108. Min Act: -0.0002

Data Index: 1552972 (Python Code)

Max Activating Token Index: 338

Text #5

Max Range: 0.1231. Min Range: -0.1231

Max Act: 0.1096. Min Act: -0.0002

Data Index: 803486 (C4 (Web Text))

Max Activating Token Index: 626

Text #6

Max Range: 0.1231. Min Range: -0.1231

Max Act: 0.1033. Min Act: -0.0002

Data Index: 579416 (C4 (Web Text))

Max Activating Token Index: 785

Text #7

Max Range: 0.1231. Min Range: -0.1231

Max Act: 0.1039. Min Act: -0.0002

Data Index: 1622303 (Python Code)

Max Activating Token Index: 29

Text #8

Max Range: 0.1231. Min Range: -0.1231

Max Act: 0.0971. Min Act: -0.0002

Data Index: 586962 (C4 (Web Text))

Max Activating Token Index: 839

Text #9

Max Range: 0.1231. Min Range: -0.1231

Max Act: 0.1026. Min Act: -0.0002

Data Index: 529783 (C4 (Web Text))

Max Activating Token Index: 355

Text #10

Max Range: 0.1231. Min Range: -0.1231

Max Act: 0.0901. Min Act: -0.0002

Data Index: 1415596 (Python Code)

Max Activating Token Index: 577

Text #11

Max Range: 0.1231. Min Range: -0.1231

Max Act: 0.0896. Min Act: -0.0002

Data Index: 572454 (C4 (Web Text))

Max Activating Token Index: 826

Text #12

Max Range: 0.1231. Min Range: -0.1231

Max Act: 0.0949. Min Act: -0.0002

Data Index: 98466 (C4 (Web Text))

Max Activating Token Index: 605

Text #13

Max Range: 0.1231. Min Range: -0.1231

Max Act: 0.1024. Min Act: -0.0002

Data Index: 442812 (C4 (Web Text))

Max Activating Token Index: 636

Text #14

Max Range: 0.1231. Min Range: -0.1231

Max Act: 0.0921. Min Act: -0.0002

Data Index: 857505 (C4 (Web Text))

Max Activating Token Index: 610

Text #15

Max Range: 0.1231. Min Range: -0.1231

Max Act: 0.0902. Min Act: -0.0002

Data Index: 392498 (C4 (Web Text))

Max Activating Token Index: 542

Text #16

Max Range: 0.1231. Min Range: -0.1231

Max Act: 0.0893. Min Act: -0.0002

Data Index: 690384 (C4 (Web Text))

Max Activating Token Index: 605

Text #17

Max Range: 0.1231. Min Range: -0.1231

Max Act: 0.0872. Min Act: -0.0002

Data Index: 540165 (C4 (Web Text))

Max Activating Token Index: 339

Text #18

Max Range: 0.1231. Min Range: -0.1231

Max Act: 0.0889. Min Act: -0.0002

Data Index: 1363223 (Python Code)

Max Activating Token Index: 236

Text #19

Max Range: 0.1231. Min Range: -0.1231

Max Act: 0.0876. Min Act: -0.0002

Data Index: 1556637 (Python Code)

Max Activating Token Index: 716