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

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

Neuron 2020 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: 0.2089. Min Range: -0.2089

Max Act: 0.2089. Min Act: -0.0003

Data Index: 969726 (C4 (Web Text))

Max Activating Token Index: 180

Text #1

Max Range: 0.2089. Min Range: -0.2089

Max Act: 0.2030. Min Act: -0.0003

Data Index: 1133158 (C4 (Web Text))

Max Activating Token Index: 32

Text #2

Max Range: 0.2089. Min Range: -0.2089

Max Act: 0.2025. Min Act: -0.0003

Data Index: 1045939 (C4 (Web Text))

Max Activating Token Index: 559

Text #3

Max Range: 0.2089. Min Range: -0.2089

Max Act: 0.1876. Min Act: -0.0003

Data Index: 928159 (C4 (Web Text))

Max Activating Token Index: 651

Text #4

Max Range: 0.2089. Min Range: -0.2089

Max Act: 0.1842. Min Act: -0.0004

Data Index: 686280 (C4 (Web Text))

Max Activating Token Index: 37

Text #5

Max Range: 0.2089. Min Range: -0.2089

Max Act: 0.1885. Min Act: -0.0003

Data Index: 510130 (C4 (Web Text))

Max Activating Token Index: 203

Text #6

Max Range: 0.2089. Min Range: -0.2089

Max Act: 0.1846. Min Act: -0.0003

Data Index: 387286 (C4 (Web Text))

Max Activating Token Index: 54

Text #7

Max Range: 0.2089. Min Range: -0.2089

Max Act: 0.1862. Min Act: -0.0003

Data Index: 1280377 (C4 (Web Text))

Max Activating Token Index: 32

Text #8

Max Range: 0.2089. Min Range: -0.2089

Max Act: 0.1859. Min Act: -0.0003

Data Index: 406080 (C4 (Web Text))

Max Activating Token Index: 39

Text #9

Max Range: 0.2089. Min Range: -0.2089

Max Act: 0.1841. Min Act: -0.0003

Data Index: 478699 (C4 (Web Text))

Max Activating Token Index: 82

Text #10

Max Range: 0.2089. Min Range: -0.2089

Max Act: 0.1810. Min Act: -0.0004

Data Index: 884218 (C4 (Web Text))

Max Activating Token Index: 380

Text #11

Max Range: 0.2089. Min Range: -0.2089

Max Act: 0.1843. Min Act: -0.0003

Data Index: 472313 (C4 (Web Text))

Max Activating Token Index: 100

Text #12

Max Range: 0.2089. Min Range: -0.2089

Max Act: 0.1796. Min Act: -0.0003

Data Index: 886006 (C4 (Web Text))

Max Activating Token Index: 335

Text #13

Max Range: 0.2089. Min Range: -0.2089

Max Act: 0.1872. Min Act: -0.0004

Data Index: 1318827 (C4 (Web Text))

Max Activating Token Index: 180

Text #14

Max Range: 0.2089. Min Range: -0.2089

Max Act: 0.1814. Min Act: -0.0004

Data Index: 1260860 (C4 (Web Text))

Max Activating Token Index: 432

Text #15

Max Range: 0.2089. Min Range: -0.2089

Max Act: 0.1778. Min Act: -0.0003

Data Index: 888849 (C4 (Web Text))

Max Activating Token Index: 858

Text #16

Max Range: 0.2089. Min Range: -0.2089

Max Act: 0.1786. Min Act: -0.0004

Data Index: 323254 (C4 (Web Text))

Max Activating Token Index: 955

Text #17

Max Range: 0.2089. Min Range: -0.2089

Max Act: 0.1949. Min Act: -0.0003

Data Index: 883716 (C4 (Web Text))

Max Activating Token Index: 574

Text #18

Max Range: 0.2089. Min Range: -0.2089

Max Act: 0.1744. Min Act: -0.0003

Data Index: 877015 (C4 (Web Text))

Max Activating Token Index: 41

Text #19

Max Range: 0.2089. Min Range: -0.2089

Max Act: 0.1765. Min Act: -0.0003

Data Index: 350556 (C4 (Web Text))

Max Activating Token Index: 700