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

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

Neuron 4897 in Layer 1

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.0940. Min Range: -0.0940

Max Act: 0.0940. Min Act: -0.0003

Data Index: 484004 (C4 (Web Text))

Max Activating Token Index: 217

Text #1

Max Range: 0.0940. Min Range: -0.0940

Max Act: 0.0875. Min Act: -0.0003

Data Index: 763076 (C4 (Web Text))

Max Activating Token Index: 895

Text #2

Max Range: 0.0940. Min Range: -0.0940

Max Act: 0.0863. Min Act: -0.0003

Data Index: 530200 (C4 (Web Text))

Max Activating Token Index: 710

Text #3

Max Range: 0.0940. Min Range: -0.0940

Max Act: 0.0862. Min Act: -0.0002

Data Index: 977955 (C4 (Web Text))

Max Activating Token Index: 983

Text #4

Max Range: 0.0940. Min Range: -0.0940

Max Act: 0.0809. Min Act: -0.0003

Data Index: 16580 (C4 (Web Text))

Max Activating Token Index: 85

Text #5

Max Range: 0.0940. Min Range: -0.0940

Max Act: 0.0810. Min Act: -0.0002

Data Index: 1355617 (C4 (Web Text))

Max Activating Token Index: 260

Text #6

Max Range: 0.0940. Min Range: -0.0940

Max Act: 0.0800. Min Act: -0.0002

Data Index: 1449087 (Python Code)

Max Activating Token Index: 544

Text #7

Max Range: 0.0940. Min Range: -0.0940

Max Act: 0.0792. Min Act: -0.0003

Data Index: 867334 (C4 (Web Text))

Max Activating Token Index: 686

Text #8

Max Range: 0.0940. Min Range: -0.0940

Max Act: 0.0772. Min Act: -0.0003

Data Index: 591628 (C4 (Web Text))

Max Activating Token Index: 798

Text #9

Max Range: 0.0940. Min Range: -0.0940

Max Act: 0.0792. Min Act: -0.0003

Data Index: 1653422 (Python Code)

Max Activating Token Index: 45

Text #10

Max Range: 0.0940. Min Range: -0.0940

Max Act: 0.0779. Min Act: -0.0003

Data Index: 660837 (C4 (Web Text))

Max Activating Token Index: 66

Text #11

Max Range: 0.0940. Min Range: -0.0940

Max Act: 0.0783. Min Act: -0.0003

Data Index: 1068296 (C4 (Web Text))

Max Activating Token Index: 160

Text #12

Max Range: 0.0940. Min Range: -0.0940

Max Act: 0.0773. Min Act: -0.0003

Data Index: 670854 (C4 (Web Text))

Max Activating Token Index: 51

Text #13

Max Range: 0.0940. Min Range: -0.0940

Max Act: 0.0770. Min Act: -0.0003

Data Index: 1136000 (C4 (Web Text))

Max Activating Token Index: 47

Text #14

Max Range: 0.0940. Min Range: -0.0940

Max Act: 0.0753. Min Act: -0.0003

Data Index: 48349 (C4 (Web Text))

Max Activating Token Index: 368

Text #15

Max Range: 0.0940. Min Range: -0.0940

Max Act: 0.0753. Min Act: -0.0002

Data Index: 1092301 (C4 (Web Text))

Max Activating Token Index: 36

Text #16

Max Range: 0.0940. Min Range: -0.0940

Max Act: 0.0761. Min Act: -0.0003

Data Index: 962396 (C4 (Web Text))

Max Activating Token Index: 24

Text #17

Max Range: 0.0940. Min Range: -0.0940

Max Act: 0.0783. Min Act: -0.0003

Data Index: 1613628 (Python Code)

Max Activating Token Index: 722

Text #18

Max Range: 0.0940. Min Range: -0.0940

Max Act: 0.0756. Min Act: -0.0003

Data Index: 1215246 (C4 (Web Text))

Max Activating Token Index: 268

Text #19

Max Range: 0.0940. Min Range: -0.0940

Max Act: 0.0731. Min Act: -0.0003

Data Index: 900519 (C4 (Web Text))

Max Activating Token Index: 347