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

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

Neuron 656 in Layer 0

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.0324. Min Range: -0.0324

Max Act: 0.0324. Min Act: -0.0006

Data Index: 556200 (C4 (Web Text))

Max Activating Token Index: 969

Text #1

Max Range: 0.0324. Min Range: -0.0324

Max Act: 0.0122. Min Act: -0.0005

Data Index: 653934 (C4 (Web Text))

Max Activating Token Index: 925

Text #2

Max Range: 0.0324. Min Range: -0.0324

Max Act: 0.0110. Min Act: -0.0005

Data Index: 1647089 (Python Code)

Max Activating Token Index: 674

Text #3

Max Range: 0.0324. Min Range: -0.0324

Max Act: 0.0108. Min Act: -0.0005

Data Index: 1055714 (C4 (Web Text))

Max Activating Token Index: 994

Text #4

Max Range: 0.0324. Min Range: -0.0324

Max Act: 0.0092. Min Act: -0.0005

Data Index: 264699 (C4 (Web Text))

Max Activating Token Index: 920

Text #5

Max Range: 0.0324. Min Range: -0.0324

Max Act: 0.0079. Min Act: -0.0005

Data Index: 1242716 (C4 (Web Text))

Max Activating Token Index: 924

Text #6

Max Range: 0.0324. Min Range: -0.0324

Max Act: 0.0077. Min Act: -0.0005

Data Index: 1620940 (Python Code)

Max Activating Token Index: 999

Text #7

Max Range: 0.0324. Min Range: -0.0324

Max Act: 0.0075. Min Act: -0.0006

Data Index: 453458 (C4 (Web Text))

Max Activating Token Index: 898

Text #8

Max Range: 0.0324. Min Range: -0.0324

Max Act: 0.0076. Min Act: -0.0005

Data Index: 2000 (C4 (Web Text))

Max Activating Token Index: 907

Text #9

Max Range: 0.0324. Min Range: -0.0324

Max Act: 0.0073. Min Act: -0.0006

Data Index: 686722 (C4 (Web Text))

Max Activating Token Index: 752

Text #10

Max Range: 0.0324. Min Range: -0.0324

Max Act: 0.0072. Min Act: -0.0005

Data Index: 71015 (C4 (Web Text))

Max Activating Token Index: 712

Text #11

Max Range: 0.0324. Min Range: -0.0324

Max Act: 0.0070. Min Act: -0.0005

Data Index: 1380412 (Python Code)

Max Activating Token Index: 888

Text #12

Max Range: 0.0324. Min Range: -0.0324

Max Act: 0.0065. Min Act: -0.0006

Data Index: 1136220 (C4 (Web Text))

Max Activating Token Index: 931

Text #13

Max Range: 0.0324. Min Range: -0.0324

Max Act: 0.0061. Min Act: -0.0004

Data Index: 1560403 (Python Code)

Max Activating Token Index: 976

Text #14

Max Range: 0.0324. Min Range: -0.0324

Max Act: 0.0063. Min Act: -0.0006

Data Index: 1230632 (C4 (Web Text))

Max Activating Token Index: 778

Text #15

Max Range: 0.0324. Min Range: -0.0324

Max Act: 0.0064. Min Act: -0.0005

Data Index: 1404319 (Python Code)

Max Activating Token Index: 899

Text #16

Max Range: 0.0324. Min Range: -0.0324

Max Act: 0.0063. Min Act: -0.0005

Data Index: 1307484 (C4 (Web Text))

Max Activating Token Index: 814

Text #17

Max Range: 0.0324. Min Range: -0.0324

Max Act: 0.0061. Min Act: -0.0005

Data Index: 1096403 (C4 (Web Text))

Max Activating Token Index: 850

Text #18

Max Range: 0.0324. Min Range: -0.0324

Max Act: 0.0062. Min Act: -0.0006

Data Index: 698916 (C4 (Web Text))

Max Activating Token Index: 897

Text #19

Max Range: 0.0324. Min Range: -0.0324

Max Act: 0.0062. Min Act: -0.0005

Data Index: 1468395 (Python Code)

Max Activating Token Index: 941