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

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

Neuron 258 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: 0.0801. Min Range: -0.0801

Max Act: 0.0799. Min Act: -0.0002

Data Index: 1287380 (C4 (Web Text))

Max Activating Token Index: 359

Text #1

Max Range: 0.0801. Min Range: -0.0801

Max Act: 0.0801. Min Act: -0.0002

Data Index: 258800 (C4 (Web Text))

Max Activating Token Index: 223

Text #2

Max Range: 0.0801. Min Range: -0.0801

Max Act: 0.0769. Min Act: -0.0002

Data Index: 171020 (C4 (Web Text))

Max Activating Token Index: 201

Text #3

Max Range: 0.0801. Min Range: -0.0801

Max Act: 0.0717. Min Act: -0.0002

Data Index: 1466740 (Python Code)

Max Activating Token Index: 177

Text #4

Max Range: 0.0801. Min Range: -0.0801

Max Act: 0.0725. Min Act: -0.0002

Data Index: 264176 (C4 (Web Text))

Max Activating Token Index: 636

Text #5

Max Range: 0.0801. Min Range: -0.0801

Max Act: 0.0594. Min Act: -0.0002

Data Index: 857550 (C4 (Web Text))

Max Activating Token Index: 701

Text #6

Max Range: 0.0801. Min Range: -0.0801

Max Act: 0.0745. Min Act: -0.0002

Data Index: 1601490 (Python Code)

Max Activating Token Index: 275

Text #7

Max Range: 0.0801. Min Range: -0.0801

Max Act: 0.0679. Min Act: -0.0002

Data Index: 1193479 (C4 (Web Text))

Max Activating Token Index: 389

Text #8

Max Range: 0.0801. Min Range: -0.0801

Max Act: 0.0738. Min Act: -0.0002

Data Index: 1377014 (Python Code)

Max Activating Token Index: 287

Text #9

Max Range: 0.0801. Min Range: -0.0801

Max Act: 0.0697. Min Act: -0.0002

Data Index: 1629088 (Python Code)

Max Activating Token Index: 254

Text #10

Max Range: 0.0801. Min Range: -0.0801

Max Act: 0.0658. Min Act: -0.0002

Data Index: 517918 (C4 (Web Text))

Max Activating Token Index: 598

Text #11

Max Range: 0.0801. Min Range: -0.0801

Max Act: 0.0680. Min Act: -0.0002

Data Index: 674951 (C4 (Web Text))

Max Activating Token Index: 1010

Text #12

Max Range: 0.0801. Min Range: -0.0801

Max Act: 0.0679. Min Act: -0.0002

Data Index: 1364679 (Python Code)

Max Activating Token Index: 854

Text #13

Max Range: 0.0801. Min Range: -0.0801

Max Act: 0.0688. Min Act: -0.0002

Data Index: 1552023 (Python Code)

Max Activating Token Index: 52

Text #14

Max Range: 0.0801. Min Range: -0.0801

Max Act: 0.0711. Min Act: -0.0002

Data Index: 1457055 (Python Code)

Max Activating Token Index: 385

Text #15

Max Range: 0.0801. Min Range: -0.0801

Max Act: 0.0715. Min Act: -0.0002

Data Index: 1493704 (Python Code)

Max Activating Token Index: 691

Text #16

Max Range: 0.0801. Min Range: -0.0801

Max Act: 0.0639. Min Act: -0.0002

Data Index: 390875 (C4 (Web Text))

Max Activating Token Index: 806

Text #17

Max Range: 0.0801. Min Range: -0.0801

Max Act: 0.0687. Min Act: -0.0002

Data Index: 1463857 (Python Code)

Max Activating Token Index: 930

Text #18

Max Range: 0.0801. Min Range: -0.0801

Max Act: 0.0672. Min Act: -0.0002

Data Index: 1536536 (Python Code)

Max Activating Token Index: 27

Text #19

Max Range: 0.0801. Min Range: -0.0801

Max Act: 0.0661. Min Act: -0.0002

Data Index: 1469341 (Python Code)

Max Activating Token Index: 874