Model: SoLU Model: 3 Layers, 2048 Neurons per Layer

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

Neuron 128 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-3l')

Text #0

Max Range: 0.0828. Min Range: -0.0828

Max Act: 0.0788. Min Act: -0.0003

Data Index: 1595243 (Python Code)

Max Activating Token Index: 227

Text #1

Max Range: 0.0828. Min Range: -0.0828

Max Act: 0.0828. Min Act: -0.0002

Data Index: 739922 (C4 (Web Text))

Max Activating Token Index: 981

Text #2

Max Range: 0.0828. Min Range: -0.0828

Max Act: 0.0766. Min Act: -0.0003

Data Index: 1376861 (Python Code)

Max Activating Token Index: 265

Text #3

Max Range: 0.0828. Min Range: -0.0828

Max Act: 0.0726. Min Act: -0.0002

Data Index: 1038291 (C4 (Web Text))

Max Activating Token Index: 468

Text #4

Max Range: 0.0828. Min Range: -0.0828

Max Act: 0.0740. Min Act: -0.0002

Data Index: 400396 (C4 (Web Text))

Max Activating Token Index: 947

Text #5

Max Range: 0.0828. Min Range: -0.0828

Max Act: 0.0762. Min Act: -0.0004

Data Index: 1049811 (C4 (Web Text))

Max Activating Token Index: 1023

Text #6

Max Range: 0.0828. Min Range: -0.0828

Max Act: 0.0729. Min Act: -0.0003

Data Index: 1387636 (Python Code)

Max Activating Token Index: 503

Text #7

Max Range: 0.0828. Min Range: -0.0828

Max Act: 0.0724. Min Act: -0.0003

Data Index: 264568 (C4 (Web Text))

Max Activating Token Index: 219

Text #8

Max Range: 0.0828. Min Range: -0.0828

Max Act: 0.0690. Min Act: -0.0003

Data Index: 158571 (C4 (Web Text))

Max Activating Token Index: 713

Text #9

Max Range: 0.0828. Min Range: -0.0828

Max Act: 0.0688. Min Act: -0.0002

Data Index: 612871 (C4 (Web Text))

Max Activating Token Index: 926

Text #10

Max Range: 0.0828. Min Range: -0.0828

Max Act: 0.0701. Min Act: -0.0003

Data Index: 1161740 (C4 (Web Text))

Max Activating Token Index: 893

Text #11

Max Range: 0.0828. Min Range: -0.0828

Max Act: 0.0696. Min Act: -0.0002

Data Index: 198112 (C4 (Web Text))

Max Activating Token Index: 950

Text #12

Max Range: 0.0828. Min Range: -0.0828

Max Act: 0.0725. Min Act: -0.0003

Data Index: 1389782 (Python Code)

Max Activating Token Index: 390

Text #13

Max Range: 0.0828. Min Range: -0.0828

Max Act: 0.0680. Min Act: -0.0003

Data Index: 1361739 (Python Code)

Max Activating Token Index: 362

Text #14

Max Range: 0.0828. Min Range: -0.0828

Max Act: 0.0700. Min Act: -0.0003

Data Index: 200665 (C4 (Web Text))

Max Activating Token Index: 735

Text #15

Max Range: 0.0828. Min Range: -0.0828

Max Act: 0.0676. Min Act: -0.0003

Data Index: 1119686 (C4 (Web Text))

Max Activating Token Index: 851

Text #16

Max Range: 0.0828. Min Range: -0.0828

Max Act: 0.0684. Min Act: -0.0002

Data Index: 1280708 (C4 (Web Text))

Max Activating Token Index: 231

Text #17

Max Range: 0.0828. Min Range: -0.0828

Max Act: 0.0656. Min Act: -0.0004

Data Index: 620844 (C4 (Web Text))

Max Activating Token Index: 983

Text #18

Max Range: 0.0828. Min Range: -0.0828

Max Act: 0.0679. Min Act: -0.0003

Data Index: 1071845 (C4 (Web Text))

Max Activating Token Index: 82

Text #19

Max Range: 0.0828. Min Range: -0.0828

Max Act: 0.0633. Min Act: -0.0002

Data Index: 409444 (C4 (Web Text))

Max Activating Token Index: 585